[Cooker] MDK10 networking feature add
Some of my users have laptops with both wired and wireless network connections. While at their desk (for example) they want to use the wired port, but when they are at lunch or in a meeting they want to use the wireless network. To rig this up, so it was dummy proof, I have both cards configured at all times with each card providing a default route. This can easily be done by using the CONFIG_IP_ROUTE_MULTIPATH kernel option. However, in order for this multi-path stuff to actually work some scripts from mandrake had to be altered. (Last laptop I configured was around mdk8.1 so there is no point in me including diffs for the scripts) Here is the minimal stuff that needs to be tweeked for 2 default static routes to work. for example, if was 192.168.1.23 and it's gw was 192.168.1.1 and was 10.1.2.23 and it's gw was 10.1.2.1 and we wanted seemless switching between the two. When starting up an interface the gateway needs to be set by /sbin/ip... NOT by ifconfig. When adding the default routes it must be done like this: /sbin/ip route append default via equalize /sbin/ip route append default via equalize in /etc/sysctl.conf: net.ipv4.ip_forward = 1 net.ipv4.conf.default.rp_filter = 0 net.ipv4.route.gc_timeout = 5 This allows seemless switching from wired and wireless networking without any interaction from the user. The gc_timeout keeps the routing cache very fresh. If this is not done it could take hours for the new connections to fail over to the alternate network connection (when you unplug the wire and walk to a meeting down the hall you don't want to wait 2 hours for the cache to flush so you can fire up your browser). To timeout existing connections faster than normal you can tweek some extra stuff... net.ipv4.tcp_keepalive_time = 30 net.ipv4.tcp_keepalive_probes = 3 net.ipv4.tcp_keepalive_intvl = 10 I'm sure I forgot to mention something... if anyone gives this a try and has problems I'd be happy to help! -- Bryan Whitehead SysAdmin - JPL - Interferometry and Large Optical Systems Phone: 818 354 2903 [EMAIL PROTECTED]
Re: [Cooker] FHS 2.3 (fwd)
[snip]
My issue is that I really don't like typing extra keys unless really
necessary. One of the nice things about linux used to be that you could
hit two keys per directory level and get quite far. Now I am going to have
to type:
/sb[TAB]/ifco[TAB] instead of /s[TAB]/ifco[TAB].
oh crap... I didn't even think of that till you mentioned it...
I know what you mean tho. I took over a solaris server and the previous
SA used /user for "user data" that jacked up my 'n for /usr .
Please don't pollute / any more (one of the reasons I hate Windows is
polluted root directories due to lack of structure by most users ...).
Exactly. Var is already a mess, lets not turn / into a mess also...
Well, /mnt/cdrom is normally temporary, as is /mnt/floppy. And again, this
is increasing the tpying I have to do (even in a graphical file manager
like Konqueror typing is quite a bit faster to get to a directory you know
the location of) if /mnt is retained and /media added.
If someone really wants a "scratch" mount area, they can make /mnt/tmp or
similar (this is what I do, usually for loopback mounts).
However, /media is a lot more descriptive than /mnt, but then it looks
quite out of place in /, being the longest directory name now
It's a toss up, I don't care one way or another personally I create
a new directory for a temp mount then remove it when i'm done
everytime... With a team of SA's I don't want to mount over /mnt when
another had already has a mount there... and checking if a directory has
already been mounted gets old real fast... ;)
If /media is used instead of /mnt I'll just delete /mnt to keep /
cleaner
BTW, we mostly have non-system data in /home/{users,groups,projects,cvs},
but we are quite small ...
we keep our data in /export/{home[1,2,3...],proj[1,2,3]} then autofs
manages /home and /proj with real names via NIS. From the users
prospective all 130+ of our unix/linux boxes look the same in terms of
data location as the only look in /home or /proj...
--
Bryan Whitehead
SysAdmin - JPL - Interferometry and Large Optical Systems
Phone: 818 354 2903
[EMAIL PROTECTED]
Re: [Cooker] FHS 2.3 (fwd)
One reason the proposal came up for /srv is that the fhs is commonly asked by sysadmins for guidance as to where they should put data for services. On linux at least, /var/www and /home/www is common, though neither seem to fit into the definitions for those areas. Is it so hard to broaden the definition of /var then? The structure under /srv has been left undefined, with some suggestions as a non-normative rationale. Being that the structure under /srv is left undefined we still end up with the same mess. /srv/database/postgresql or /srv/postgresql or /src/disk1/postgresql This does not solve a problem: /srv will still be a mess from distro to distro. Or, even worse, some distros may ignore /srv sticking to the previous LSB/FHS (undefined) waiting for clarification or changes in future versions of the FHS. So we'll end up with both /var /home /srv depending on the flavor of linux. No, the rationale states that data of interest to a specific user should go in that user's home directory. Here is the full text of the rationale: "This main purpose of specifying this is so that users may find the location of the data files for particular service, and so that services which require a single tree for readonly data, writable data and scripts (such as cgi scripts) can be reasonably placed. Data that is only of interest to a specific user should go in that users home directory. The methodology used to name subdirectories of /srv is unspecified as there is currently no consensus on how this should be done. One method for structuring data under /srv is by protocol, eg. ftp, rsync, www, and cvs. On large systems it can be useful to structure /srv by administrative context, such as /srv/physics/www, /srv/compsci/cvs, etc. This setup will differ from host to host. Therefore, no program should rely on these locations." This assumes data is arranged and stored in seprate places based off of services. rsync, cvs, ftp, and www can easily share the same target data. Yup, we don't have FHS police :-) Admins do make local changes. In this case its more about setting up defaults for applications/packages to place or access their data. Then please keep it out of / . If this is just to be a dumping ground for apps to leave their default junk we don't need to clutter up / with another dir do we? /var seems to have always been the ugly stepchild of unwanted crap... I'm fine with adding more to it... ;) The /media entry doesn't bother me... but I'd rather see them redefine /mnt to be what most use it for anyway: exactly what they propose /media does. I can already visualize my inbox filling up for the mdk10 upgrade, "/mnt/cdrom is GONE!!! please FIX" At the moment on linux we have /mnt/cdrom, /cdrom and /media/cdrom used. The main objection to /mnt/cdrom has been the older unix tradition of using /mnt as a temporary mount kind of scratch area. Though I do wonder how much this really matters these days. Chris -- Bryan Whitehead SysAdmin - JPL - Interferometry and Large Optical Systems Phone: 818 354 2903 [EMAIL PROTECTED]
Re: [Cooker] FHS 2.3 (fwd)
Just to recap. I think redhat is right on this one... / should not get more messy than it already is... my original reply is below that was sent to cooker (in case you don't subscribe). Bryan Whitehead wrote: Stew Benedict wrote: OK folks. FHS 2.3 is currently struggling with a couple of controversial proposals. They would like Mandrake's opinion, as part of what FHS tries to do is formalize current convention. United Linux based distros use these currently. My take is Red Hat is against them. I'm not sure how to reply with a formal Mandrake position, but perhaps posting here will give me more of a feel on how the community views these proposals. Since you asked... My opinion is /srv is kinda lame idea to begin with. Where we put our data for services seems to be a personal preference. At JPL many use /export/home[1,2,3...] for nfs/samba home directories. Project space would get /export/proj[1,2,3...]. A database would get /export/data. Others use /data/ or /disk[1,2,3...] or . In other words, this change looks like they want to enforce a certain way of locating / arranging data for various services. I say leave this out... this pulls us away from how other UNIX's deal with this problem: allow the SA to choose his own scheme. Also the description seems pretty stupid, "/srv - data generated by users for the services the system offers". Can't this also mean data in /home? if the system offers samba as a service should the data now go in /srv ? What about nfs exported home directories do they belong in /srv now? The intention of unifying the location of where data is kept is a semi-good idea for those who need a decision made for them, but it will be largly ignored by shops who already have a scheme in place and don't want to have thier linux machines differ from their Solaris/AIX/IRIX machines. I don't see why keeping the "default" in /var is bad as many distro's do now. At least it doesn't clutter up my / with yet-another entry. The /media entry doesn't bother me... but I'd rather see them redefine /mnt to be what most use it for anyway: exactly what they propose /media does. I can already visualize my inbox filling up for the mdk10 upgrade, "/mnt/cdrom is GONE!!! please FIX" -- Bryan Whitehead SysAdmin - JPL - Interferometry and Large Optical Systems Phone: 818 354 2903 [EMAIL PROTECTED]
Re: [Cooker] FHS 2.3 (fwd)
Stew Benedict wrote: OK folks. FHS 2.3 is currently struggling with a couple of controversial proposals. They would like Mandrake's opinion, as part of what FHS tries to do is formalize current convention. United Linux based distros use these currently. My take is Red Hat is against them. I'm not sure how to reply with a formal Mandrake position, but perhaps posting here will give me more of a feel on how the community views these proposals. Since you asked... My opinion is /srv is kinda lame idea to begin with. Where we put our data for services seems to be a personal preference. At JPL many use /export/home[1,2,3...] for nfs/samba home directories. Project space would get /export/proj[1,2,3...]. A database would get /export/data. Others use /data/ or /disk[1,2,3...] or . In other words, this change looks like they want to enforce a certain way of locating / arranging data for various services. I say leave this out... this pulls us away from how other UNIX's deal with this problem: allow the SA to choose his own scheme. Also the description seems pretty stupid, "/srv - data generated by users for the services the system offers". Can't this also mean data in /home? if the system offers samba as a service should the data now go in /srv ? What about nfs exported home directories do they belong in /srv now? The intention of unifying the location of where data is kept is a semi-good idea for those who need a decision made for them, but it will be largly ignored by shops who already have a scheme in place and don't want to have thier linux machines differ from their Solaris/AIX/IRIX machines. I don't see why keeping the "default" in /var is bad as many distro's do now. At least it doesn't clutter up my / with yet-another entry. The /media entry doesn't bother me... but I'd rather see them redefine /mnt to be what most use it for anyway: exactly what they propose /media does. I can already visualize my inbox filling up for the mdk10 upgrade, "/mnt/cdrom is GONE!!! please FIX" -- Bryan Whitehead SysAdmin - JPL - Interferometry and Large Optical Systems Phone: 818 354 2903 [EMAIL PROTECTED]
[Cooker] kernel oops for 2.4.19-35mdkenterprise
I attached the oops. I emailed security as this is a security releated release kernel... -- Bryan Whitehead SysAdmin - JPL - Interferometry and Large Optical Systems Phone: 818 354 2903 [EMAIL PROTECTED] Unable to handle kernel paging request at virtual address 000e4030 printing eip: c015e1c9 *pde = Oops: CPU:1 EIP:0010:[iput+57/704]Not tainted EIP:0010:[]Not tainted EFLAGS: 00010202 eax: 000e4020 ebx: f05f3ca0 ecx: f05f3cb0 edx: e34829e0 esi: 000e4020 edi: d01cd000 ebp: f7edfecc esp: f7edfeb4 ds: 0018 es: 0018 ss: 0018 Process kswapd (pid: 5, stackpage=f7edf000) Stack: e4cc5b40 f6fd0640 f8acaa26 e34829e0 f05f3ca0 c029f280 f7edfee4 c015b7a2 f05f3ca0 f05f3ca0 e34829e0 000e f7edfefc c015cad2 e34829e0 dd9b45a0 00de 000e f7edff0c c015bb57 e34829e0 7aab f7edff18 c015bdb7 Call Trace:[] [dentry_iput+98/128] [prune_one_dentry+50/152] [prune_dcache+103/160] [shrink_dcache_memory+39/64] Call Trace:[] [] [] [] [] [shrink_cache+632/1232] [shrink_caches+73/96] [try_to_free_pages+98/272] [kswapd_balance_pgdat+110/176] [kswapd_balance+40/64] [kswapd+157/192] [] [] [] [] [] [] [kswapd+0/192] [_stext+0/96] [arch_kernel_thread+46/64] [kswapd+0/192] [] [] [] [] Code: 8b 46 10 85 c0 0f 85 5c 02 00 00 c7 44 24 04 1c 00 2a c0 8d <1>Unable to handle kernel paging request at virtual address 2c313438 printing eip: c015e1c9 *pde = Oops: CPU:0 EIP:0010:[iput+57/704]Not tainted EIP:0010:[]Not tainted EFLAGS: 00010206 eax: 2c313428 ebx: f7636740 ecx: f7636750 edx: cebae9e0 esi: 2c313428 edi: f4183000 ebp: f6ce7b88 esp: f6ce7b70 ds: 0018 es: 0018 ss: 0018 Process nfsd (pid: 1318, stackpage=f6ce7000) Stack: ccdadaa0 d8944180 f8acaa26 cebae9e0 f7636740 c029f280 f6ce7ba0 c015b7a2 f7636740 f7636740 cebae9e0 001e f6ce7bb8 c015cad2 cebae9e0 d22305c0 016d 001e f6ce7bc8 c015bb57 cebae9e0 89df f6ce7bd4 c015bdb7 Call Trace:[] [dentry_iput+98/128] [prune_one_dentry+50/152] [prune_dcache+103/160] [shrink_dcache_memory+39/64] Call Trace:[] [] [] [] [] [shrink_cache+632/1232] [shrink_caches+73/96] [try_to_free_pages+98/272] [balance_classzone+116/400] [__alloc_pages+372/640] [af_packet:__insmod_af_packet_O/lib/modules/2.4.19-35mdkenterprise/ker+-815648/96] [] [] [] [] [] [] [page_cache_read+149/240] [generic_file_readahead+199/352] [do_generic_file_read+823/1152] [file_read_actor+0/256] [generic_file_read+154/368] [file_read_actor+0/256] [] [] [] [] [] [] [] [af_packet:__insmod_af_packet_O/lib/modules/2.4.19-35mdkenterprise/ker+-814673/96] [] [af_packet:__insmod_af_packet_O/lib/modules/2.4.19-35mdkenterprise/ker+-829482/96] [] [] [] [] [] [] [] [] [af_packet:__insmod_af_packet_O/lib/modules/2.4.19-35mdkenterprise/ker+-761760/96] [] [] [] [] [] [] [] [] [] [] [] [] [] [] [] [arch_kernel_thread+46/64] [] [] [] [] [] [] []
Re: [Cooker] mdk9.2 rc2 kickstart problem
Pixel wrote: Bryan Whitehead <[EMAIL PROTECTED]> writes: rc2 seems to have problems installing packages listed in the auto_inst.cfg file in the default_packages section. I list autofs, allong with a huge list of other packages, but they do not get installed. is the label 'default_packages' now wrong? It's like the kickstart ignores the list of packages I give it... it works nicely here. Maybe you can provide the /root/drakx/report.bug.gz? Will send ASAP... -- Bryan Whitehead SysAdmin - JPL - Interferometry and Large Optical Systems Phone: 818 354 2903 [EMAIL PROTECTED]
[Cooker] mdk9.2 rc2 kickstart problem
rc2 seems to have problems installing packages listed in the auto_inst.cfg file in the default_packages section. I list autofs, allong with a huge list of other packages, but they do not get installed. is the label 'default_packages' now wrong? It's like the kickstart ignores the list of packages I give it... -- Bryan Whitehead SysAdmin - JPL - Interferometry and Large Optical Systems Phone: 818 354 2903 [EMAIL PROTECTED]
Re: [Cooker] Re: What happened to am-utils?
Buchan Milne wrote: Bryan Whitehead wrote: Thanks for the kick in the pants. Looks like autofs can do what I need. Looks like this "feature" slipped in during 9.0 (or 8.2) and I didn't notice. Now if only autofs would support direct mounts we'd be another step closer to Solaris... ;) Could you explain what it is/how it works? auto_direct allows you to assign a directory a specific disk/nfsserver when it is accessed. this means you can make any directory managed by autofs without having autofs control the parent directory. most common use of direct mount is /var/mail. This config wants "mail" in /var/ to be mounted from a nfs server, but you can't have "/var" managed by autofs. So autofs will watch the "mail" directory directly. When a program enters/opens the directory autofs will mount the directory from it's configured source. An auto.direct entry would look like this: /var/mail -actimeo=0 s383:/export/mail someone may say... "why don't you just keep in mounted??!?!". Because of stale NFS handles when servers go down. Even when the server comes back up the client may be hosed... if you have several hundred workstations this can suck real fast. NFS mounts should only be mounted when "in use". If no one is looking at /var/mail then /var/mail should not be mounted via nfs. Currently on linux we would have to change /var/mail to be a symlink to /net/s383/export/mail. However, this becomes a problem when applications decide to save the "mail folder" as /net/s383/export/mail. If we move our mail services to another server we'll have problems. But mail is just a simple stupid example... /usr/local is software that is installed by the SA. Many shops share this over nfs (very common at JPL) so each workstation can have new software by installing it on the server... auto_direct would look like this: /usr/local -rw s383:/export/opt/local having /usr/local as a symlink to /net/export/opt/local can screw up many things. For example, if a project compiles software on their workstation configure/makefiles and binarys can end up with /net/export/opt/local in them... resulting in huge problems when servers are moved, or decommissioned. -- Bryan Whitehead SysAdmin - JPL - Interferometry and Large Optical Systems Phone: 818 354 2903 [EMAIL PROTECTED]
Re: [Cooker] Re: What happened to am-utils?
As you will see, the only entries in our LDAP automount maps are those we
need all the time (/home/{users,groups,projects}) on most machines (the
ones people need to be sure of, so we can modify the LDAP automount
entries and be sure each machine will use the correct share on next
access), I have done no configuration on /net and /misc (just using the
default auto.master, auto.net, auto.misc - LDAP stuff is picked up
automagically without *any* configuration besides that done during
installation for LDAP auth).
I can browse any machine our firewall will let me use NFS/portmap to ...
We aren't a "big unix shop" (we have more Windows boxen), but I still
haven't seen anything am-utils can do that autofs can't.
Regards,
Buchan
Thanks for the kick in the pants. Looks like autofs can do what I need.
Looks like this "feature" slipped in during 9.0 (or 8.2) and I didn't
notice.
Now if only autofs would support direct mounts we'd be another step
closer to Solaris... ;)
--
Bryan Whitehead
SysAdmin - JPL - Interferometry and Large Optical Systems
Phone: 818 354 2903
[EMAIL PROTECTED]
Re: [Cooker] Re: What happened to am-utils?
Buchan Milne wrote: Bryan Whitehead wrote: David Walser wrote: Bryan Whitehead wrote: Bryan Whitehead wrote: am-utils is a very much-needed package for Linux to integrate into big "UNIX shops". NASA/JPL (at least in my section) is a big "UNIX Shop". Wiothout am-utils we are pretty screwed... is there a reason this package was removed from 9.2 ? Has it just been renamed? BTW, this package was removed from 9.1 also. I made my own RPM from the only in 9.0, but it's still a hassle. When I reccomend Mandrake to other admins the first thing they ask is "Where fsck is am-utils"? I then send them some RPM's... But it would be nice if it was shipped standard... And you are using it instead of autofs because...? Because autofs is very limited. Example: [EMAIL PROTECTED] bgmilne]$ cd /net/hermes [EMAIL PROTECTED] hermes]$ ls home/ does hermes only export one mountpoint? Here is a typical machine at JPL: [EMAIL PROTECTED] ~]$ cd /net/s383 [EMAIL PROTECTED] s383]$ ls export/ [EMAIL PROTECTED] s383]$ ls export/ home1/ home6/ proj1/ proj14/ proj2/ proj24/ proj29/ proj5/ home2/ home7/ proj10/ proj15/ proj20/ proj25/ proj3/ proj6/ home3/ home8/ proj11/ proj17/ proj21/ proj26/ proj30/ proj7/ home4/ mail/ proj12/ proj18/ proj22/ proj27/ proj31/ proj8/ home5/ opt/proj13/ proj19/ proj23/ proj28/ proj4/ proj9/ [EMAIL PROTECTED] s383]$ home1, home2, etc etc are all different disk arrays and mount points with different export options to different machines. [EMAIL PROTECTED] hermes]$ cd /net/hercules [EMAIL PROTECTED] hercules]$ ls home/ [EMAIL PROTECTED] hercules]$ ls home/ groups/ projects/ users/ [EMAIL PROTECTED] hercules]$ cd ../ftp [EMAIL PROTECTED] ftp]$ ls var/ (using the default autofs config, plus our own maps in LDAP which handle /cae - ordinarily I don't ever go in /net since most stuff I need is under /cae and linked into /home, so these were automounted by autofs) Thanks for showing me the world of "every machine has a LDAP entry". But I live in the world of NIS and DNS name resolution; with the addition of possible exports that are associated with neither... The defualt config also has some things under /misc (cdrom,floppy). BTW, if you choose LDAP auth during install, you should now get autofs installed, and if you have LDAP automount maps setup well, they should work out-the-box (I still need to test this though ...). Maybe David Walser can comment on some of these actions would be worthwhile for NIS also (although it may be too late). If you can show me how I can get this working with autofs (with NIS or without NIS)... Then I'll eat crow and shutup... Now lets say a computer outside your org (not in your LDAP, or NIS) exports NFS to you; can you do this: cd /net// and browse? What about an IP? [EMAIL PROTECTED] ~]$ cd /net/137.78.61.94/ [EMAIL PROTECTED] 137.78.61.94]$ ls export/ [EMAIL PROTECTED] 137.78.61.94]$ With am-utils I can browse any machine in the world that allows me to NFS mount regardless of it's entry in NIS or LDAP or DNS or "autofs map"... -- Bryan Whitehead SysAdmin - JPL - Interferometry and Large Optical Systems Phone: 818 354 2903 [EMAIL PROTECTED]
Re: [Cooker] What happened to am-utils?
To recap (without the bitching about uucp and postfix) autofs is crap compaired to am-utils. If we can only have one, ditch autofs as am-utils can do all that autofs can do plus much needed MORE. -- Bryan Whitehead SysAdmin - JPL - Interferometry and Large Optical Systems Phone: 818 354 2903 [EMAIL PROTECTED]
Re: [Cooker] Re: What happened to am-utils?
But it's in contrib, there is an alternative in main => what's the problem? The problem is while I'm a Mandrake monkey and can handle the misisng package other SA's who just want a fully shipped distro don't want to have to 1) ask for it 2) look for it 3) get it 4) install it. They want basic things like nfs-network browsing to WORK out of the box we CANNOT do that on mandrake as it ships. I have to add this as an add-on post-install. Also, since I added it make the list of crap I have to "watch for security related problems" bigger. Well, if am-utils is better, it should replace autofs in main (autofs can go to contrib). But, AFAIK, there is no alternative for uucp, and packages in main (postfix) contain entries in the default config for it (thus you kind of expect it to be available ...). When mandrake had them both I used them both (I still use them both but I have to add am-utils myself) I used autofs for NIS mappings. But since It's limited to having maps for mounting, I needed am-utils for on the fly dynamic nfs browsing. If autofs was removed I'd use am-utils for both... but autofs can't do both... so I have to add am-utils. BTW, I haven't used am-utils, only autofs, but does anyone know if am-utils has ldapv3 support (autofs at present only has ldapv2 support, and this makes connections to openldap-2.1.x not work out-the-box). If mandrake ditched autofs for am-utils I'd be happy and just use that. -- Bryan Whitehead SysAdmin - JPL - Interferometry and Large Optical Systems Phone: 818 354 2903 [EMAIL PROTECTED]
Re: [Cooker] Re: What happened to am-utils?
David Walser wrote: Bryan Whitehead wrote: Bryan Whitehead wrote: am-utils is a very much-needed package for Linux to integrate into big "UNIX shops". NASA/JPL (at least in my section) is a big "UNIX Shop". Wiothout am-utils we are pretty screwed... is there a reason this package was removed from 9.2 ? Has it just been renamed? BTW, this package was removed from 9.1 also. I made my own RPM from the only in 9.0, but it's still a hassle. When I reccomend Mandrake to other admins the first thing they ask is "Where fsck is am-utils"? I then send them some RPM's... But it would be nice if it was shipped standard... And you are using it instead of autofs because...? Because autofs is very limited. Example: With am-utils I can easily "browse" my entire network's NFS exports. cd /net// and I'm browsing the exports of a machine -- Bryan Whitehead SysAdmin - JPL - Interferometry and Large Optical Systems Phone: 818 354 2903 [EMAIL PROTECTED]
Re: [Cooker] What happened to am-utils?
Bryan Whitehead wrote: am-utils is a very much-needed package for Linux to integrate into big "UNIX shops". NASA/JPL (at least in my section) is a big "UNIX Shop". Wiothout am-utils we are pretty screwed... is there a reason this package was removed from 9.2 ? Has it just been renamed? BTW, this package was removed from 9.1 also. I made my own RPM from the only in 9.0, but it's still a hassle. When I reccomend Mandrake to other admins the first thing they ask is "Where fsck is am-utils"? I then send them some RPM's... But it would be nice if it was shipped standard... -- Bryan Whitehead SysAdmin - JPL - Interferometry and Large Optical Systems Phone: 818 354 2903 [EMAIL PROTECTED]
[Cooker] What happened to am-utils?
am-utils is a very much-needed package for Linux to integrate into big "UNIX shops". NASA/JPL (at least in my section) is a big "UNIX Shop". Wiothout am-utils we are pretty screwed... is there a reason this package was removed from 9.2 ? Has it just been renamed? -- Bryan Whitehead SysAdmin - JPL - Interferometry and Large Optical Systems Phone: 818 354 2903 [EMAIL PROTECTED]
Re: [Cooker] Missing monitor in XFdrake database
here is another: Apple; Apple 23" Cinema HD; APP9218; 30.0-90.0; 50.0-70.0; 1 To get the full use of the monitor this modline is needed in X: Section "Modes" identifier "APP9218" Modeline "1920x1200" 155.0 1920 1984 2016 2144 1200 1203 1206 1212 EndSection Is there a database for modlines? Thanks again for the help in getting this added to Mandrake. -- Bryan Whitehead SysAdmin - JPL - Interferometry and Large Optical Systems Phone: 818 354 2903 [EMAIL PROTECTED]
Re: [Cooker] Missing monitor in XFdrake database
Thierry Vignaud wrote: Bryan Whitehead <[EMAIL PROTECTED]> writes: IBM; IBM T85A; IBM252f; 30.0-82.0; 56.0-75.0; 1 just added in cvs Thanks! :) I will submit more that are missing. (Got many flatpanels that are not working). you're welcome :-) Here is another: NEC; NEC LCD2080UX; NEC6605; 30.0-96.0; 48.0-87.0; 1 Thanks again for the help. -- Bryan Whitehead SysAdmin - JPL - Interferometry and Large Optical Systems Phone: 818 354 2903 [EMAIL PROTECTED]
Re: [Cooker] Missing monitor in XFdrake database
could you try adding the following line "IBM; IBM Analog Flatpanel; IBM252f; 30.0-82.0; 56.0-75.0; 1" to /usr/share/ldetect-lst/MonitorsDB ? format is "VendorName; ModelName; EISA_ID; HorizSync_Range; VertRefresh_Range; dpms_support?" you can get your monitor's EISA ID by running the following command: ddcxinfos | fgrep -i eisa i suppose this is how you get IBM252f ? the 1 at end of line must be kept *only* if your monitor support dpms for power saving. put the proper name instead of "IBM Analog Flatpanel" since i do not think this is what would be the model name. if you send me back the right line, i'll add support for your monitor in the monitor database IBM; IBM T85A; IBM252f; 30.0-82.0; 56.0-75.0; 1 I will submit more that are missing. (Got many flatpanels that are not working). -- Bryan Whitehead SysAdmin - JPL - Interferometry and Large Optical Systems Phone: 818 354 2903 [EMAIL PROTECTED]
Re: [Cooker] Missing monitor in XFdrake database
Thierry Vignaud wrote: Bryan Whitehead <[EMAIL PROTECTED]> writes: here is another: Apple; Apple 23" Cinema HD; APP9218; 30.0-90.0; 50.0-70.0; 1 To get the full use of the monitor this modline is needed in X: Section "Modes" identifier "APP9218" Modeline "1920x1200" 155.0 1920 1984 2016 2144 1200 1203 1206 1212 EndSection pixel ? Even without the modline, can the monitor be added to the database? Then at least 1600x1200 will work out of the box. :) Thanks Is there a way to get cvs write access to commit this stuff myself? or should I just keep feeding them thru you/cooker? -- Bryan Whitehead SysAdmin - JPL - Interferometry and Large Optical Systems Phone: 818 354 2903 [EMAIL PROTECTED]
[Cooker] Missing monitor in XFdrake database
I have an IBM Analog Flatpanel monitor that is missing from the monitor database, or XFdrake can't configure the monitor ddcxinfos says it's a "IBM252f". The vert and horizontal ranges are: HorizSync 30.0-82.0 VertRefresh 56.0-75.0 Max resolution is 1280x1024x32 (I use 1280x1024x24) Can this monitor be added to the XFdrake database so it works in kickstart? Thanks. -- Bryan Whitehead SysAdmin - JPL - Interferometry and Large Optical Systems Phone: 818 354 2903 [EMAIL PROTECTED]
Re: [Cooker] state of kickstart Auto X config
Buchan Milne wrote: On Fri, 15 Aug 2003, Bryan Whitehead wrote: Who at mandrake do I need to contact to get this working? Most likely [EMAIL PROTECTED] I've been given the go-ahead to help hack X / kickstart so we can get this working. I just need to know who i contact for getting this into kickstart and X... (Also where to start... is it just Kickstart that isn't rebust enough... or is X the problem?) I think all the tools (DrakX / XFdrake) use the same library code, so either it's that (if XFdrake doesn't do the right thing if you take all the defaults) or it's DrakX (if XFdrake does the right thing but the kickstart stuff doesn't), or the ddxinfo is wrong, which you can tell by looking at the output of 'ddcxinfos'. Regards, Buchan Looks like ddcxinfos is working. The machines that are not being configed correctly do have full ddc info... Not only that but a huge list of modelines. I guess somthing needs to be done in XFdrake / Drakx. This helps with the Analog side... what about digital flat panels? Do they send resolution info the same way? Thanks for the pointer, I now see the problem isn't with the monitor not providing the needed info - it's kickstart not using the info. Now I can concentrate on patches to get kickstart fully automated for X! -- Bryan Whitehead SysAdmin - JPL - Interferometry and Large Optical Systems Phone: 818 354 2903 [EMAIL PROTECTED]
[Cooker] state of kickstart Auto X config
Currently I'm running into problems with Kickstart unable to automagically configure my XF86Config-4 files. After kickstart I have to manually hack them (or write automated scripts) to fix them. This is becomming a huge problem as our linux user base grows. Although we try to get the same hardware/brand To keep things easier we still run into problems. It seems that kickstart isn't using the DDC information from X correctly, or X isn't giving the correct DDC information. This results in X being configured for a 800x600 monitor (safe mode default config?). Also, when using digital flatpanel screens. It never sets up X correctly. Who at mandrake do I need to contact to get this working? I've been given the go-ahead to help hack X / kickstart so we can get this working. I just need to know who i contact for getting this into kickstart and X... (Also where to start... is it just Kickstart that isn't rebust enough... or is X the problem?) -- Bryan Whitehead SysAdmin - JPL - Interferometry and Large Optical Systems Phone: 818 354 2903 [EMAIL PROTECTED]
[Cooker] DVI graphics card work with Mandrake?
I'm looking for a graphics card that will work out of the box with mandrake 9.1. In otherwords, I don't want to download some binary driver, or rebuild X, or do some weird kernel hack The cards I've been looking at are these: ATI Radeon VE VGA and DVI ATI FIRE GL X1 1-2 VGA and DVI nVidia QuadroFX 500 VGA and DVI nVidia QuadroFX 1000 VGA and DVI nVidia Quadro NVS 280 VGA and DVI The card will be driving a 20" flatpanel UltraSharp 2000FP from Dell. I don't really have a problem with hacking the XF86Config-4 file, but I don't like having to hack in external drivers... Do any of these cards have opensource/xfree drivers that ship with mandrake that are "good enough" ? BTW, They do not need todo fancy 3D/OpenGL, all the apps will be basic 2D stuff... -- Bryan Whitehead SysAdmin - JPL - Interferometry and Large Optical Systems Phone: 818 354 2903 [EMAIL PROTECTED]
Re: [Cooker] Bug: Horde/Imp & php >= 4.3.0
David Walser wrote: horde/imp don't use a db backend. turba does, but we horde/imp DO use a db backend http://www.horde.org/faq/admin/print.php#p13 You can configure to not save anything between sessions so it doesn't use a database but then horde/imp becomes much less usefull. don't even have it configured to work out of the box, so if you configured it so that it actually works, you'd know you needed to install mysql or pgsql. --- magic <[EMAIL PROTECTED]> wrote: Also of note: I believe there should be an added require for the rpms. Either php-mysql or php-pgsql (depending on which db is used on the backend) should be required. (That was a different error that took awhile to figure out, and non discriptive php error messages didn't help). Thanks, S __ Do you Yahoo!? SBC Yahoo! DSL - Now only $29.95 per month! http://sbc.yahoo.com -- Bryan Whitehead SysAdmin - JPL - Interferometry and Large Optical Systems Phone: 818 354 2903 [EMAIL PROTECTED]
Re: [tlyons@mandrakesoft.com: More bad perl requires (was Re: [Cooker]emacs-EPL-0.7-3mdk.i586.rpm)]
Todd Lyons wrote: -BEGIN PGP SIGNED MESSAGE- Hash: SHA1 I originally sent this to an internal list, but then I realized contribs were probably handled by others. These need to be fixed. Blue skies... Todd This is a bug in rpm (rpm-build to be exact). find.requires is the problem, located in /usr/lib/rpm. look at line 165. When you build you "own" perl based rpm it read the site config version as 5.8.00 but then writes the requires as 5.800 which breaks perl builds... Pain in the ass I've been dealing with for my perl modules I build at JPL. - - Forwarded message from Todd Lyons <[EMAIL PROTECTED]> - Date: Fri, 14 Feb 2003 10:31:01 -0800 Subject: More bad perl requires (was Re: [Cooker] emacs-EPL-0.7-3mdk.i586.rpm) Pixel wrote on Thu, Feb 13, 2003 at 10:31:47PM +0100 : Charles A Edwards <[EMAIL PROTECTED]> writes: emacs-EPL-0.7-3mdk.i586.rpm is uninstallable as it builds with a require for "perl-base-5.800" bad bad package. fixing. [root@fiji /work/Mandrake-devel/contrib/i586]# for i in [pP]erl-*; do rpm -qp --requires $i | grep "5\.800" && echo $i; done perl-base >= 5.800 Perl-Tidy-20020425-3mdk.noarch.rpm perl-base >= 5.800 perl-Cache-Cache-1.01-3mdk.noarch.rpm perl-base >= 5.800 perl-NKF-1.71-3mdk.i586.rpm perl-base >= 5.800 perl-Net-Netmask-1.9002-2mdk.i586.rpm perl-base >= 5.800 perl-SNMP-MIB-Compiler-0.06-2mdk.noarch.rpm perl-base >= 5.800 perl-SOAP-Lite-0.55-2mdk.noarch.rpm perl >= 5.800 perl-Text-Kakasi-1.05-2mdk.i586.rpm Blue skies... Todd - -- | MandrakeSoft USA | Security is like an onion. It's made | | http://www.mandrakesoft.com | made up of several layers and makes | | http://www.mandrakelinux.com | you cry. --Howard Chu| Mandrake Cooker Devel Version, Kernel 2.4.21pre4-5mdk - - End forwarded message - - -- Todd Lyons -- MandrakeSoft, Inc. http://www.mandrakesoft.com/ Favourite shell: bash, though I also like 'init=/usr/bin/emacs' --Andrew Tridgell Mandrake Cooker Devel Version, Kernel 2.4.21pre4-5mdk -BEGIN PGP SIGNATURE- Version: GnuPG v1.2.1 (GNU/Linux) iD8DBQE+TTablp7v05cW2woRArBQAJ40hfh8Xvo5WdO53LvX0v9TUrY8KQCfQLxB 8bJrOAbZdXGCoqHKFcsahOI= =M7pM -END PGP SIGNATURE- -- Bryan Whitehead SysAdmin - JPL - Interferometry Systems and Technology Phone: 818 354 2903 [EMAIL PROTECTED]
Re: [Cooker] Please don't make urpmi stupid
Steve Fox wrote: On Wed, 2003-01-08 at 12:32, Bryan Whitehead wrote: In the real world, sysadmins just find a solution (like redoing the crap ass RPM someone made) and move on with life. If you don't like dependancy checking, then don't use urpmi. Use the plain rpm command. Write your own scripts as a wrapper. Or just patch urpmi yourself for your own needs. It's not rocket-science! ;) Wow, what a well thought-out response! You're a freaking genius! I NEED dependency checking. What I am asking for is the tool to only worry about dependencies which are affected by the action being requested. Thanks for clarifying. The message I replied to originally implied you didn't like the dependacy checking / resolving features of urpmi. There is no reason for urpmi to care about unrelated dependencies which I made a conscious decision to break knowing very well the potential results. Your right on this. I second the motion to keep urpmi from being some wannabe apt. -- Bryan Whitehead SysAdmin - JPL - Interferometry Systems and Technology Phone: 818 354 2903 [EMAIL PROTECTED]
Re: [Cooker] Please don't make urpmi stupid
Steve Fox wrote: On Mon, 2003-01-06 at 23:19, Vox wrote: I actually like to think that urpmi getting old unresolved deps flagged is A Good Thing(tm). If you don't care for deps, don't use a package *manager*. I totally appreciate that for packages which are being upgraded/installed. But for stuff that's not being modified, it shouldn't worry about it. The whole "all dependencies must always be resolved" is nice if you live in your closed little world where smart packagers like Mandrake create proper dependencies. But in the Real World (tm), there are idiots who package things to fit in their little world. They won't listen to me when I say to use a "Requires: java" instead of "Requires: IBM-JDK". In the real world, sysadmins just find a solution (like redoing the crap ass RPM someone made) and move on with life. If you don't like dependancy checking, then don't use urpmi. Use the plain rpm command. Write your own scripts as a wrapper. Or just patch urpmi yourself for your own needs. It's not rocket-science! ;) Just don't advocate breaking a tool other people like for your *small* need of getting one damn broken rpm to smoothly work! Find a better tool for your needs or create your own tools from scratch. -- Bryan Whitehead SysAdmin - JPL - Interferometry Systems and Technology Phone: 818 354 2903 [EMAIL PROTECTED]
Re: [Cooker] MS Fonts in a bag
Murray J. Root wrote: On Tue, Dec 17, 2002 at 04:22:34PM -0800, J. Greenlees wrote: Ben Reser wrote: On Wed, Dec 11, 2002 at 10:03:42AM -0500, Austin Acton wrote: PLEASE let's not start the should we/shouldn't we debate again. ~snip~ Didn't last long: http://avi.alkalay.net/software/msfonts/download/ Anyone have a mirror? the site is dead. well, since when you buy a computer you invariably get windows [ except the mandrake pc's from the mandrake store ], we all have the font's available, and can import them from windows, without breaking any license or laws. my personal collection of ms tt fonts is over 1500 fonts. of course I am importing them to my mandrake boxes. big noise about getting something we all already have. You're making an assumption that isn't even close to valid. Not everyone buys prebuilt boxes with OS installed. There are VERY MANY people who don't. Windows does not come with mainboards (yet), so these people do NOT have MS Fonts already. Although it is generally trivial to get them - Ben's RPM, for example. -- Bryan Whitehead SysAdmin - JPL - Interferometry Systems and Technology Phone: 818 354 2903 [EMAIL PROTECTED]
Re: [Cooker] XFS+HIGHMEM lockup, with fix.
Todd Lyons wrote: -BEGIN PGP SIGNED MESSAGE- Hash: SHA1 Bryan Whitehead wrote on Mon, Nov 18, 2002 at 04:52:57PM -0800 : ahh, great work, just came home after being away for the weekend, and that one made me very happy=) I'd hope Mandrake would release a new kernel. If we are hitting this bug, I'm sure others are Bryan, is there any indication that this bug exists in only the XFS driver (ie patches we got were bad...grabbed at just the wrong moment in time) or is it more kernel wide? It seems to be isolated to the XFS driver only in my limited view. this only is in the XFS driver. The patch I sent only will affect the xfs module. Blue skies... Todd - -- MandrakeSoft USA http://www.mandrakesoft.com Mandrake: An amalgam of good ideas from RedHat, Debian, and MandrakeSoft. All in all, IMHO, an unbeatable combination. --Levi Ramsey on Cooker ML Cooker Version mandrake-release-9.1-0.1mdk Kernel 2.4.19-19mdksecure -BEGIN PGP SIGNATURE- Version: GnuPG v1.2.1 (GNU/Linux) iD8DBQE93TwQlp7v05cW2woRAuBpAJ9NHqeydQNuqoKqFVFmz20jqsRycwCgsDdZ ka7WVDJe/HbgCdrv4SW5Sdg= =Ik2f -END PGP SIGNATURE- -- Bryan Whitehead SysAdmin - JPL - Interferometry Systems and Technology Phone: 818 354 2903 [EMAIL PROTECTED]
Re: [Cooker] XFS+HIGHMEM lockup, with fix.
Per Øyvind Karlsen wrote: Bryan Whitehead wrote: After working with SGI for the past week, with kdb backtraces and many rebuilds of the kernel. The problem with the xfs filesystem as mandrake ships it was found. Finally! Now I can use all the memory we bought for our machines!! :) I included a patch that needs to be dropped into the 2.4.19-q16/patches directory in the kernel source as shipped with mandrake 9.0. Please apply and release a new kernel... :) diff -ru linux-2.4.19/fs/xfs.orig/pagebuf/page_buf_io.c linux-2.4.19/fs/xfs/pagebuf/page_buf_io.c --- linux-2.4.19/fs/xfs.orig/pagebuf/page_buf_io.c2002-11-13 17:25:21.0 -0800 +++ linux-2.4.19/fs/xfs/pagebuf/page_buf_io.c2002-11-13 17:31:40.0 -0800 @@ -309,7 +309,8 @@ __pb_block_prepare_write_async(ip, page, cpoff, cpoff+csize, at_eof, NULL, pbmapp, PBF_WRITE); -memset((void *) (kmap(page) + cpoff), 0, csize); +/* __pb_block_prepare_write already kmap'd the page */ +memset((void *) (page_address(page) + cpoff), 0, csize); pagebuf_commit_write_core(ip, page, cpoff, cpoff + csize); pos = ((loff_t)page->index << PAGE_CACHE_SHIFT) + cpoff + csize; Only in linux-2.4.19/fs/xfs/pagebuf: page_buf_io.c.orig ahh, great work, just came home after being away for the weekend, and that one made me very happy=) I'd hope Mandrake would release a new kernel. If we are hitting this bug, I'm sure others are -- Bryan Whitehead SysAdmin - JPL - Interferometry Systems and Technology Phone: 818 354 2903 [EMAIL PROTECTED]
Re: [Cooker] XFS+HIGHMEM lockup, with fix.
Sweet! I'm impressed that you found such an obscure bug. Could you go through some of the steps that you went through to do this debugging? I would suggest the following mentionables: 1) enabling kdb Not that hard in the mandrake kernel, just switch the kdb option in the .spec file from 0 to 1. 2) interpreting the data I did a full backtrace of every process that was on the machine. (After the machine locked up). once your in the kdb console you just type "ps" for the process list, and then "bta" to backtrace all processes. The problem was there was NO xfs backtraces, however one processes was screwed up and a backtrace could not be done. This ment that something in that process got totally hosed. it's also the process that hard locked the machine. I pointed out that the load on the machine can be very low when the bug is triggered, and that it doesn't get triggered when there is no high memory available. (even when running the highmem kernel) In linux, when HIGHMEM is turned on there is still "regular" (LOWMEM) memory. As long as xfs was only playing with the LOWMEM segment of memory no lockups. But as soon as xfs used HIGHMEM (no matter the load ont he machine) it locked hard. 3) What made you zero in on that particular memset (ties back to #2) I ran SGI's pre3 of XFS 1.2 kernels and had no problems. After that I used thier own brew of 1.1 with highmem, and didn't have a problem. After I told SGI they sent me a one liner patch. 4) Documentation that says the __pb_block_prepare_write already kmap'ed the page. If the answer is "use the source Luke" then so be it SGI read the source... and knows it... ;) Mandrake must have pulled the xfs patches from SGI at exactly the wrong time :( -- Bryan Whitehead SysAdmin - JPL - Interferometry Systems and Technology Phone: 818 354 2903 [EMAIL PROTECTED]
[Cooker] XFS+HIGHMEM lockup, with fix.
After working with SGI for the past week, with kdb backtraces and many rebuilds of the kernel. The problem with the xfs filesystem as mandrake ships it was found. Finally! Now I can use all the memory we bought for our machines!! :) I included a patch that needs to be dropped into the 2.4.19-q16/patches directory in the kernel source as shipped with mandrake 9.0. Please apply and release a new kernel... :) -- Bryan Whitehead SysAdmin - JPL - Interferometry Systems and Technology Phone: 818 354 2903 [EMAIL PROTECTED] diff -ru linux-2.4.19/fs/xfs.orig/pagebuf/page_buf_io.c linux-2.4.19/fs/xfs/pagebuf/page_buf_io.c --- linux-2.4.19/fs/xfs.orig/pagebuf/page_buf_io.c 2002-11-13 17:25:21.0 -0800 +++ linux-2.4.19/fs/xfs/pagebuf/page_buf_io.c 2002-11-13 17:31:40.0 -0800 @@ -309,7 +309,8 @@ __pb_block_prepare_write_async(ip, page, cpoff, cpoff+csize, at_eof, NULL, pbmapp, PBF_WRITE); - memset((void *) (kmap(page) + cpoff), 0, csize); + /* __pb_block_prepare_write already kmap'd the page */ + memset((void *) (page_address(page) + cpoff), 0, csize); pagebuf_commit_write_core(ip, page, cpoff, cpoff + csize); pos = ((loff_t)page->index << PAGE_CACHE_SHIFT) + cpoff + csize; Only in linux-2.4.19/fs/xfs/pagebuf: page_buf_io.c.orig
Re: [Cooker] kernel-enterprise hard lock on 9.0rc2
Per Øyvind Karlsen wrote: nonono, on the server it runs 2.4.19-16mdk, I just brought this old post to live again;) The problem still exist in 2.4.19-enterprise-mdk and I have'nt seen anything in the changelog of the latest release about it, and besides if it's fixed, it's such a serious bug that there should be an update for 9.0 available(when it's fixed) agreed. I got a trace from using the kernel-console-serial link and sent it awile ago... Still no work from mandrake. :( the trace and stuff is at huey.jpl.nasa.gov/~driver Brent Hasty wrote: On Friday 08 November 2002 05:03 pm, Per Øyvind Karlsen wrote: can you download MDK 9.0 full release, it has updates that might help your problems with the RC-2 you are using. Per Øyvind Karlsen wrote: Brad Felmey wrote: On Wed, 2002-09-18 at 13:22, Bryan Whitehead wrote: Testing one of our 2GB Dual-Pentium III (1ghz) machines we get a hard lock with kernel enterprise. kernel-enterprise-2.4.19.9mdk-1-1mdk.i586.rpm The lock up seems to be related to the scsi driver. aic7xxx. The last message that is echoed to the syslog consol is: kernel: (scsi0:A:0:0): Locking max tag count at 64 after that the machine is dead. No ping, but keyboard works enough for Alt-SysRq stuff. Only problem is Alt-SysRq keys won't work. :( syncing the disk results in the machine "trying" (It starts to sync the disk) but will never finish, there is no HD activity. Then the keyboard will not respond anymore. Trying to get a process list, or memory stats will finish off the lock also. (as in it will become so useless that reset button is only thing working). If my first sysrq key is to reboot it will. I ran memtest yesterday and all memory is ok. Mandrake 8.1 has no problems on this machine. Heavy disk activity triggers the lock, like installing an RPM. :) It will always hard lock when installing the "FlashPlayer-5.0-8mdk" RPM. any suggestions? or specific sysrq keys you'd like me to try in locked up state? (there is no oops or anything like that on the consol window) I have a dual-1GHz PIII Coppermine system with 2GB on a Supermicro Serverworks board and Adaptec 3200S that has this exact problem. Only with highmem/kernel-enterprise after 2.4.8. RPM will kill the box exactly as you've described. I've had to retrograde the box and stay away from current kernels. I reported a bug some days earlier about problems with the enterprise kernel, but not the same scsi driver(mine is sym53c8xx) And as I said this did'nt happen with the old 2.4.5 kernel(from mdk-cooker) it was running for almost a year, but now it's had a mem upgrade and is running the kernel-enterprise kernel, and we've experienced lockups with 2.4.18 and 2.4.19. But this does'nt happen when using rpm, and the machine is alive, responding to ping etc. but could'nt read from the disks, and happens under high load(not very high, just compiling something could trigger it ), at first I thought it was a gcc-3.2 issue, but when trying with kernel-enterprise-2.4.18.6mdk-1-1mdk this also happened, so I'll guess this might be related Has this been fixed yet? We never got any feedback from Juan.. Don't mean to nag, but this one is a pretty critical one, and when fixed maybe a MDKA update is in place..? oh well, I'll just continue to run kernel-smp meanwhile, although lesser ram available:\ -- Bryan Whitehead SysAdmin - JPL - Interferometry Systems and Technology Phone: 818 354 2903 [EMAIL PROTECTED]
Re: [Cooker] enterprise kernel lockup - with trace
Brent Hasty wrote: On Friday 08 November 2002 03:53 pm, Bryan Whitehead wrote: what is the key sequence to show "Memory" "ShowReg" "Show State"? huey.jpl.nasa.gov/~driver/crash.txt This has the full kernel boot sequence upto and including the hard lock. After the machine locked up I Alt-SysRq'ed "Memory" "ShowReg" and "Show State". Sync will not work, it will just be stuck. This is on a machine running 2.4.19-16mdkenterprise with 2GB of RAM. Memtest shows all memory is good. This is the best I can do for logs... If this can't find the problem Me and Todd tried many differnt options being passsed to the kernel... and nothing worked. Any idea's? Hit Alt-SysRq-H for the help menu at a Linux console. -- Bryan Whitehead SysAdmin - JPL - Interferometry Systems and Technology Phone: 818 354 2903 [EMAIL PROTECTED]
[Cooker] enterprise kernel lockup - with trace
huey.jpl.nasa.gov/~driver/crash.txt This has the full kernel boot sequence upto and including the hard lock. After the machine locked up I Alt-SysRq'ed "Memory" "ShowReg" and "Show State". Sync will not work, it will just be stuck. This is on a machine running 2.4.19-16mdkenterprise with 2GB of RAM. Memtest shows all memory is good. This is the best I can do for logs... If this can't find the problem Me and Todd tried many differnt options being passsed to the kernel... and nothing worked. Any idea's? -- Bryan Whitehead SysAdmin - JPL - Interferometry Systems and Technology Phone: 818 354 2903 [EMAIL PROTECTED]
Re: [Cooker] MDK 8.2, 1024 MB RAM
Norman Carver wrote: I have a number of machines with 1.5-2GB of ram on them, and have been running the Enterprise (himem) kernels on them (starting with MDK 8.0). We have had no problems that I would blame on the Enterprise kernels. In fact, the only problems we have had were memory corruption problems with MDK 8.2, which were (apparently) related to msec. Fixed these problems by disabling msec checks. Same problem happened on a machine using the stock 8.2 kernel as well. I cannot quite recall what messages we would find in the logs with the error, but it was a kernel attempt to address memory that failed. We think it was related to the zip drives as well (and supermount). Currently am running 8.2 and 9.0 Enterprise kernels without problems. Todd from Mandrake visited and we tracked the bug down to a XFS problem. Aparently XFS+HIGHMEM = hard lock at random loads... ext2 and ext3 don't seem to have the problem at all. Only when the FS is XFS does the lockups happen. I hope they find a fix!!! :) -- Bryan Whitehead SysAdmin - JPL - Interferometry Systems and Technology Phone: 818 354 2903 [EMAIL PROTECTED]
[Cooker] pm3fb driver bug and fix
The pm3fb driver does not have all the same features when it's a module vs being compiled directly into the kernel. I alerted the author who provided a fix. (This issue was using a digital flatpanel screen was immposible because passing "flatpanel=1" to the driver as a module would not work. Only passing to the driver built into the kernel worked) This is a patch for drivers/video/pm3fb.c from the author. It would be nice if this patch is applied for future kernel updates... Thanks! -- Bryan Whitehead SysAdmin - JPL - Interferometry Systems and Technology Phone: 818 354 2903 [EMAIL PROTECTED] pm3fb.patch.gz Description: application/gzip
Re: [Cooker] MDK 8.2, 1024 MB RAM
Peter Magnusson wrote: On Wed, 23 Oct 2002, Bryan Whitehead wrote: Actaully it's more than that. A machine with 2GB of ram only sees this without highmem: total used free sharedbuffers cached Mem:904940 185524 719416 0 84 107520 -/+ buffers/cache: 77920 827020 Swap: 3084400 03084400 We can't run any HIGHMEM kernels cause they hard lock on every machine after minimal use. :( So how do you do to use more memory? Or dont you use it at all? I just don't use all my memory big waste, but constant crashing is just too much to put up with on the enterprise kernel :'( So far no word/help from Mandrake.... :( :( -- Bryan Whitehead SysAdmin - JPL - Interferometry Systems and Technology Phone: 818 354 2903 [EMAIL PROTECTED]
[Cooker] xrdb bug in mdk9.0
xrdb is broken in mandrake 9.0 [driver@mythic ~]$ xrdb sh: line 1: /lib/cpp: No such file or directory [driver@mythic ~]$ the /lib/cpp link points to an entry that doesn't exist in /etc/alternatives [driver@mythic ~]$ ls -al /lib/cpp lrwxr-xr-x1 root root 25 Oct 25 07:20 /lib/cpp -> /etc/alternatives/lib_cpp [driver@mythic ~]$ lib_cpp is missing. I changed broken symlink to point at /etc/alternatives/cpp. So now /lib/cpp points to /etc/alternatives/cpp and everything works. Please fix! :) (with this broken .Xdefaults will not work correctly unless xrdb gets -nocpp passed to it... which it doesn't) This is a pretty big bug for unix shops... many old timers (and die hard advanced X users) tweak everything with .Xdefaults... When this doesn't work they freak out. :) -- Bryan Whitehead SysAdmin - JPL - Interferometry Systems and Technology Phone: 818 354 2903 [EMAIL PROTECTED]
Re: [Cooker] Why can't I access/play audio cd's?
Benj FitzPatrick wrote: I have tried both cd drives, and both say I do not have permission to access them (logged in under my identity and root). This only happens with audio cd's, and (of course) all of the cd playing programs give me the same error. How can I fix this? Thanks. What's the error? I'd suspect supermount. Only cause I had the same problem at one point. Shuttle ak31 with ac97 sound, xp1600+, 256MB pc2100, evga geforce4 mx420, 52x cd (slave), 32x10x48 cdrw (master) __ Do you Yahoo!? Y! Web Hosting - Let the expert host your web site http://webhosting.yahoo.com/ -- Bryan Whitehead SysAdmin - JPL - Interferometry Systems and Technology Phone: 818 354 2903 [EMAIL PROTECTED]
[Cooker] foomatic mdk9.0 bug?
This is mostly a heads up. Using foomatic to install a printer from the command line has broken in 9.0 for setting the default BINDING option. In previous versions of mandrake the BINDING flag for printing was set to LONGEDGE. Now the default is SHORTEDGE. This makes full duplex printing print the "not like it used to" way. Setting binding to LONGEDGE means that each side of a paper will be printed from top to bottom. (In otherwords: think of printing a bunch of documents and binding them together on the long side of the paper - out have a book... you read left-right). With SHORTEDGE printing is done top-bottom then from bottom-top on the other side. So each side is opposite in orientation. (So imagine binding a notebook on the short side of the paper, so as you flip thru the book everything is read top-bottom). I has a script that would set up a printer like this: foomatic-configure -s lprng -c "lpd:///lp" -n -d Postscript -p HP-LaserJet_2200 On previous versions of Mandrake this would set up a printer that would set the defaults of the printer too DEPLUX=ON and BINDING=LONGEDGE. With 9.0 the defaults are DEPLUX=ON and BINDING=SHORTEDGE. I fixed my script so this is no longer a problem: foomatic-configure -s lprng -c "lpd:///lp" -n -d Postscript -p HP-LaserJet_2200 -o BINDING=LONGEDGE But it was a pain in the butt to track down Maybe this is a foomatic change and not a mandrake bug?? -- Bryan Whitehead SysAdmin - JPL - Interferometry Systems and Technology Phone: 818 354 2903 [EMAIL PROTECTED]
Re: [Cooker] MDK 8.2, 1024 MB RAM
andre wrote: On Tuesday 22 October 2002 22:59, Per Øyvind Karlsen wrote: smp and highmem, and highmem are still a little buggy, if you look at some of the posts last month, several people had problems with the highmem kernels. btw. do you really need the highmem kernel for 1024 MB ram? I thought that the kernel supported up to 1024 MB withouth the highmem kernel anyways, I dont see what your problem has to do with highmem support compiled in the standard kernel or not. you will see 50MB or so not. IMHO no problem Actaully it's more than that. A machine with 2GB of ram only sees this without highmem: total used free sharedbuffers cached Mem:904940 185524 719416 0 84 107520 -/+ buffers/cache: 77920 827020 Swap: 3084400 03084400 We can't run any HIGHMEM kernels cause they hard lock on every machine after minimal use. :( So far no word/help from Mandrake.... :( -- Bryan Whitehead SysAdmin - JPL - Interferometry Systems and Technology Phone: 818 354 2903 [EMAIL PROTECTED]
Re: [Cooker] Missing package in mandrake 9.0 breaks printing.
Well, It's not on any of the 3 CD's. :( rlpr is required for LPDng to print to other lpd print ques. Why is it not on the main disk with LPRng? Till Kamppeter wrote: > As far as I know "rlpr" is in the contribs. > > Till > > > Bryan Whitehead wrote: > >> LPRng needs rlpr too function with remote printer ques >> >> I pulled rlpr from Mandrake 8.1 to get printing working with LPRng >> >> (You can also grep the files in /etc/foomatic/lpd for rlpr) >> >> We add printers by directly calling foomatic >> >> foomatic-configure -s lprng -c "lpd://lp171-309/lp" -n lp171-309 -d >> Postscript -p HP-LaserJet_4100 >> >> after "installing" a printer as above look at your /etc/foomatic/lpd >> entry. >> >> reading foomatic-configure you'll see why rlpr is needed. (just search >> for rlpr) >> > -- Bryan Whitehead SysAdmin - JPL - Interferometry Systems and Technology Phone: 818 354 2903 [EMAIL PROTECTED]
[Cooker] Missing package in mandrake 9.0 breaks printing.
LPRng needs rlpr too function with remote printer ques I pulled rlpr from Mandrake 8.1 to get printing working with LPRng (You can also grep the files in /etc/foomatic/lpd for rlpr) We add printers by directly calling foomatic foomatic-configure -s lprng -c "lpd://lp171-309/lp" -n lp171-309 -d Postscript -p HP-LaserJet_4100 after "installing" a printer as above look at your /etc/foomatic/lpd entry. reading foomatic-configure you'll see why rlpr is needed. (just search for rlpr) -- Bryan Whitehead SysAdmin - JPL - Interferometry Systems and Technology Phone: 818 354 2903 [EMAIL PROTECTED]
Re: [Cooker] 9.0 final when ?
Brook Humphrey wrote: > On Friday 20 September 2002 10:28 am, Bryan Whitehead wrote: > >>Guillaume Cottenceau wrote: >> >>>Stéphane Teletchéa <[EMAIL PROTECTED]> writes: >>> >>>>Is it worth testing this WE and reporting monday ? >>> >>>I think the "first" final will be built tonight, so yes if >>>showstopper bugs are found this WE it will help us much. >>> >>>so please report showstopper bugs (e.g. not non important bugs). >> >>Isn't the kernel-enterprise hard-lock a showstopper bug? :( Havn't heard >>anything about it from the mandrake team. > > > Tes and Ihave a compaq proliant 850 dual cpu that 8.2 installs on just fine > that cant make it through the install process for 9.0. > Yea, but on install the enterprise kernel isn't running. Once your system is up (and has over 900mb of memory) then you'd be running that kernel. Does the install kernel have HIGHMEM on? I don't have any problems durring the install, it's only the specific enterprise kernel. Perhaps it's a combo of HIGHMEM + SMP ? -- Bryan Whitehead SysAdmin - JPL - Interferometry Systems and Technology Phone: 818 354 2903 [EMAIL PROTECTED]
Re: [Cooker] 9.0 final when ?
Guillaume Cottenceau wrote: > Stéphane Teletchéa <[EMAIL PROTECTED]> writes: > > >>Is it worth testing this WE and reporting monday ? > > > I think the "first" final will be built tonight, so yes if > showstopper bugs are found this WE it will help us much. > > so please report showstopper bugs (e.g. not non important bugs). > Isn't the kernel-enterprise hard-lock a showstopper bug? :( Havn't heard anything about it from the mandrake team. -- Bryan Whitehead SysAdmin - JPL - Interferometry Systems and Technology Phone: 818 354 2903 [EMAIL PROTECTED]
Re: [Cooker] kernel-enterprise hard lock on 9.0rc2
Randy wrote: > Bryan Whitehead wrote: > >> Randy wrote: >> [snip] >> >>> >>> Works just fine on my dual Athlon MP system running *fresh* B4: >>> >>> Kernel 2.4.19-5mdkenterprise on a Dual-processor i686 / \l >>> >>> Tiger MPX, IDE based, with Promise 133 as well. >>> >>> -Rand E >>> >> >> Do you NFS allot? > > > No, I havn't used NFS on this machine (at least not serving). > I am able to do intense hard drive activity which I thought > was related. BTW I have a 1024MB stick in it. > > I'll give RC3 a try as soon as I can. > > -Rand > I think the lock is nfs/rpc related, so that's why your not getting the lock-up. I'm building a new kernel to test... I'm probably wrong. -- Bryan Whitehead SysAdmin - JPL - Interferometry Systems and Technology Phone: 818 354 2903 [EMAIL PROTECTED]
Re: [Cooker] kernel-enterprise hard lock on 9.0rc2
Randy wrote: [snip] > > Works just fine on my dual Athlon MP system running *fresh* B4: > > Kernel 2.4.19-5mdkenterprise on a Dual-processor i686 / \l > > Tiger MPX, IDE based, with Promise 133 as well. > > -Rand E > Do you NFS allot? -- Bryan Whitehead SysAdmin - JPL - Interferometry Systems and Technology Phone: 818 354 2903 [EMAIL PROTECTED]
Re: [Cooker] kernel-enterprise hard lock on 9.0rc2
Guillaume Cottenceau wrote: > Bryan Whitehead <[EMAIL PROTECTED]> writes: > > >>Mandrake? any comment? or tips to resolve this? > > > I've forwarded to the kernel team. I don't know if Juan will have > the time to look over it.. > Can you forward this to the mandrake kernel team? I'd be willing to run tests. Does mandrake add these patches from the nfs guy? http://www.fys.uio.no/~trondmy/src/2.4.19/ there is a NFS deadlock patch for HIGHMEM systems. Might be what we are getting. specifically, linux-2.4.19-01-fix_kmap1.dif linux-2.4.19-01-fix_kmap2.dif linux-2.4.19-01-fix_kmap3.dif there are a bunch of other patches that might also help with deadlocks on HIGHMEM, but the descriptions are not clear. -- Bryan Whitehead SysAdmin - JPL - Interferometry Systems and Technology Phone: 818 354 2903 [EMAIL PROTECTED]
Re: [Cooker] Problems creating new media source
Does mandrake add these patches from the nfs guy? http://www.fys.uio.no/~trondmy/src/2.4.19/ there is a NFS deadlock patch for HIGHMEM systems. Might be what we are getting. specifically, linux-2.4.19-01-fix_kmap1.dif linux-2.4.19-01-fix_kmap2.dif linux-2.4.19-01-fix_kmap3.dif there are a bunch of other patches that might also help with deadlocks on HIGHMEM, but the descriptions are not clear. Ben Reser wrote: > On Thu, Sep 19, 2002 at 08:34:07AM -0500, SillyZ wrote: > >>urpmi.addmedia CD1 file://rpms >>added medium CD1 >>reading rpm files from [/rpms] >>unable to read rpm files from [/rpms]: bad rpm >>/rpms/imwheel-0.9.9-3mdk.i586.rpm >> >>no hdlist file found for medium "CD1" >>examining synthesis file [/var/lib/urpmi/synthesis.hdlist.CD1.cz] >>problem reading synthesis file of medium "CD1" >>unable to update medium "CD1" >> >>I have also attempted a urpmi.addmedia -f and getting the same result. >> >>Thought someone should report this. All the rpms are copied off the cdroms >>into a single dir ( I have also tried using /rpms/cd1 , /rpms/cd2 and >>/rpms/cd3 respectivly ). Produces the same error. RC2 did not error when >>preforming the same function. > > > When starting new threads please start with a new message. Don't reply > to existing messages. It messes up threading for those of us with > threaded email clients. Your message is under a thread about a kernel > problem. Myself and others have been known to just Delete entire > threads that aren't interesting. Which means your message might not be > seen by some people. > -- Bryan Whitehead SysAdmin - JPL - Interferometry Systems and Technology Phone: 818 354 2903 [EMAIL PROTECTED]
Re: [Cooker] RC3: printerdrake: lprng, not cups is default printingsubsystem
Damon Lynch wrote: > Hi Till, > > When I 1st ran printerdrake, I didn't notice that lprng was set as the > printing system. I rebooted, ran it again, and that's when I noticed. > I was wondering why the remote cups printer was not showing up!! ;-) > > Damon > After I just re-wrote all my scripts to deal with CUPS!! AH! I'm still happy to see lprng back... CUPS isn't nearly as stable. -- Bryan Whitehead SysAdmin - JPL - Interferometry Systems and Technology Phone: 818 354 2903 [EMAIL PROTECTED]
Re: [Cooker] Cooker way to much...
Yura Gusev wrote: > There are way to many messages per day. I think we should do something > about it. > > try the lkml this list feels quiet... ;) -- Bryan Whitehead SysAdmin - JPL - Interferometry Systems and Technology Phone: 818 354 2903 [EMAIL PROTECTED]
Re: [Cooker] kernel-enterprise hard lock on 9.0rc2
Bjarne Thomsen wrote: > Have you also tried the latest redhat kernel like > kernel-bigmem-2.4.18-5.i686.rpm ? > It would be nice to know if redhat has a fix. > > -- Bjarne I can try it. I'll let you know what happens > > On Wed, 2002-09-18 at 23:57, Brad Felmey wrote: > >>On Wed, 2002-09-18 at 13:22, Bryan Whitehead wrote: >> >> >>>Testing one of our 2GB Dual-Pentium III (1ghz) machines we get a hard >>>lock with kernel enterprise. >>>kernel-enterprise-2.4.19.9mdk-1-1mdk.i586.rpm >>> >>>The lock up seems to be related to the scsi driver. aic7xxx. The last >>>message that is echoed to the syslog consol is: >>>kernel: (scsi0:A:0:0): Locking max tag count at 64 >>> >>>after that the machine is dead. No ping, but keyboard works enough for >>>Alt-SysRq stuff. Only problem is Alt-SysRq keys won't work. :( syncing >>>the disk results in the machine "trying" (It starts to sync the disk) >>>but will never finish, there is no HD activity. Then the keyboard will >>>not respond anymore. >>> >>>Trying to get a process list, or memory stats will finish off the lock >>>also. (as in it will become so useless that reset button is only thing >>>working). >>> >>>If my first sysrq key is to reboot it will. >>> >>>I ran memtest yesterday and all memory is ok. Mandrake 8.1 has no >>>problems on this machine. >>> >>>Heavy disk activity triggers the lock, like installing an RPM. :) >>> >>>It will always hard lock when installing the "FlashPlayer-5.0-8mdk" RPM. >>> >>>any suggestions? or specific sysrq keys you'd like me to try in locked >>>up state? (there is no oops or anything like that on the consol window) >> >>I have a dual-1GHz PIII Coppermine system with 2GB on a Supermicro >>Serverworks board and Adaptec 3200S that has this exact problem. Only >>with highmem/kernel-enterprise after 2.4.8. RPM will kill the box >>exactly as you've described. I've had to retrograde the box and stay >>away from current kernels. >>-- >>Brad Felmey >> >> > > > -- Bryan Whitehead SysAdmin - JPL - Interferometry Systems and Technology Phone: 818 354 2903 [EMAIL PROTECTED]
Re: [Cooker] kernel-enterprise hard lock on 9.0rc2
Bjarne Thomsen wrote: > We have consistently encountered hard system locks > with all enterprise 2.4.18 and 2.4.19 kernels while > transferring relatively large amount of data (images) > either via ftp or NFS. We heavily use NFS. The RPM's I was installing were on a NFS server. > We did NOT encounter any problems with the 2.4.8 > enterprise kernel from ML 8.1, so we have been using > that kernel with ML 8.2, until somebody finds a fix. > > We did not have any problem with the ordinary 2.4.18 > or 2.4.19 kernels, so the problem is only found in > the enterprise kernel. Yep, same here. Only the 9.0 2.4.19 kernels have the problem. > As an experiment I compiled one of the 2.4.18 enterprise > kernels without the HIGHMEM option (the only change made). > This kernel did not have the lock-ups, so I concluded > that the problem had something to do with the HIGHMEM > code. > > It is actually strange to me that this problem has > been so little commented, considering the low prices on RAM. We had thought it was a hardware error so we ran a bunch of tests before we reported it. > Does anybody know if this problem has been or will be fixed > in 2.4.20 ? > > Bjarne Thomsen > Department of Physics and Astronomy > University of Aarhus > Mandrake? any comment? or tips to resolve this? > On Wed, 2002-09-18 at 20:22, Bryan Whitehead wrote: > >>Testing one of our 2GB Dual-Pentium III (1ghz) machines we get a hard >>lock with kernel enterprise. >>kernel-enterprise-2.4.19.9mdk-1-1mdk.i586.rpm >> >>The lock up seems to be related to the scsi driver. aic7xxx. The last >>message that is echoed to the syslog consol is: >>kernel: (scsi0:A:0:0): Locking max tag count at 64 >> >>after that the machine is dead. No ping, but keyboard works enough for >>Alt-SysRq stuff. Only problem is Alt-SysRq keys won't work. :( syncing >>the disk results in the machine "trying" (It starts to sync the disk) >>but will never finish, there is no HD activity. Then the keyboard will >>not respond anymore. >> >>Trying to get a process list, or memory stats will finish off the lock >>also. (as in it will become so useless that reset button is only thing >>working). >> >>If my first sysrq key is to reboot it will. >> >>I ran memtest yesterday and all memory is ok. Mandrake 8.1 has no >>problems on this machine. >> >>Heavy disk activity triggers the lock, like installing an RPM. :) >> >>It will always hard lock when installing the "FlashPlayer-5.0-8mdk" RPM. >> >>any suggestions? or specific sysrq keys you'd like me to try in locked >>up state? (there is no oops or anything like that on the consol window) >> >> >>-- >>Bryan Whitehead >>SysAdmin - JPL - Interferometry Systems and Technology >>Phone: 818 354 2903 >>[EMAIL PROTECTED] >> >> > > > -- Bryan Whitehead SysAdmin - JPL - Interferometry Systems and Technology Phone: 818 354 2903 [EMAIL PROTECTED]
[Cooker] kernel-enterprise hard lock on 9.0rc2
Testing one of our 2GB Dual-Pentium III (1ghz) machines we get a hard lock with kernel enterprise. kernel-enterprise-2.4.19.9mdk-1-1mdk.i586.rpm The lock up seems to be related to the scsi driver. aic7xxx. The last message that is echoed to the syslog consol is: kernel: (scsi0:A:0:0): Locking max tag count at 64 after that the machine is dead. No ping, but keyboard works enough for Alt-SysRq stuff. Only problem is Alt-SysRq keys won't work. :( syncing the disk results in the machine "trying" (It starts to sync the disk) but will never finish, there is no HD activity. Then the keyboard will not respond anymore. Trying to get a process list, or memory stats will finish off the lock also. (as in it will become so useless that reset button is only thing working). If my first sysrq key is to reboot it will. I ran memtest yesterday and all memory is ok. Mandrake 8.1 has no problems on this machine. Heavy disk activity triggers the lock, like installing an RPM. :) It will always hard lock when installing the "FlashPlayer-5.0-8mdk" RPM. any suggestions? or specific sysrq keys you'd like me to try in locked up state? (there is no oops or anything like that on the consol window) -- Bryan Whitehead SysAdmin - JPL - Interferometry Systems and Technology Phone: 818 354 2903 [EMAIL PROTECTED]
Re: [Cooker] good news about mozilla+dual head problems
Bryan Whitehead wrote: > Todd Lyons wrote: > >> Bryan Whitehead wrote on Tue, Sep 17, 2002 at 03:37:23PM -0700 : >> >>> Anyway, the solution is to not run a font server for X on linux, >>> instead, all the font paths need to be listed in the XF86Config-4 >>> file. This also fixed the problem with KDE "locking up" (many apps >>> did for me). >> >> >> >> I recall another instance where this exact step fixed the problems but I >> can't recall the details about it. I do not think it involved dual >> head, but I could be wrong because I've slept (a little ) since then. >> In this case, Mozilla was a red herring as the problem all along was >> X. Does Gnome exhibit this same behavior? (it should since it's >> running under the same X). >> >> Blue skies...Todd > > > I never had any problems with gnome1 or gnome2. That's why I didn't > think of trying this... > > Also KDE usually has problems for me on a single head, so I never > bothered with trying to fix KDE. (no one uses it in my userbase so I > don't really test it) > > I guess gnomecc had a problem... a user just told me that the program > never worked for him till I changed the font server settings. (Sometimes > I don't know of bugs till they are fixed... quiet userbase) > BTW, After filing a bug report with the mozilla guys they found a duplicate. The bug report along with possible fixes is here: http://bugzilla.mozilla.org/show_bug.cgi?id=136362 -- Bryan Whitehead SysAdmin - JPL - Interferometry Systems and Technology Phone: 818 354 2903 [EMAIL PROTECTED]
Re: [Cooker] autofs + NIS problem
David Walser wrote: > Problem: > Even if you don't have nis in the automount: line in > /etc/nsswitch.conf, automount uses NIS. This is standard behavior. (NIS on Solaris and other Unix's do this) This way one can have a nis server export an auto.master file so one doesn't have to modify every client machine on the network when a change to the auto.master file happens. If your nis server does not export an auto.master file then you won't have a problem. If it does, but you need a different auto.master file for your autofs on linux, then remove nis and nisplus from nsswitch.conf. Autofs will still work with nis/nisplus/yp maps but it will not autoload the auto.master from your nis server. This should not be changed, then nis on linux would be further different from other unix's. > Reason: > in /etc/init.d/autofs function getrawmounts() on line > 121 you see: > nis*|yp) > > which doesn't distinguish between nis and nisplus > (script is probably being lazy, because later when it > does the ypcat, it doesn't make a difference). I'm > not sure what you could do to tell whether you're > connected to NIS or NIS+, but if there's something, > the script should do that and be smarter about what it > does. > > Since you (Juan) probably didn't write this script, if > you want me to report this somewhere else, just let me know. > > ______ > Do you Yahoo!? > Yahoo! News - Today's headlines > http://news.yahoo.com -- Bryan Whitehead SysAdmin - JPL - Interferometry Systems and Technology Phone: 818 354 2903 [EMAIL PROTECTED]
Re: [Cooker] good news about mozilla+dual head problems
Todd Lyons wrote: > Bryan Whitehead wrote on Tue, Sep 17, 2002 at 03:37:23PM -0700 : > >>Anyway, the solution is to not run a font server for X on linux, >>instead, all the font paths need to be listed in the XF86Config-4 file. >>This also fixed the problem with KDE "locking up" (many apps did for me). > > > I recall another instance where this exact step fixed the problems but I > can't recall the details about it. I do not think it involved dual > head, but I could be wrong because I've slept (a little ) since then. > In this case, Mozilla was a red herring as the problem all along was > X. Does Gnome exhibit this same behavior? (it should since it's > running under the same X). > > Blue skies... Todd I never had any problems with gnome1 or gnome2. That's why I didn't think of trying this... Also KDE usually has problems for me on a single head, so I never bothered with trying to fix KDE. (no one uses it in my userbase so I don't really test it) I guess gnomecc had a problem... a user just told me that the program never worked for him till I changed the font server settings. (Sometimes I don't know of bugs till they are fixed... quiet userbase) -- Bryan Whitehead SysAdmin - JPL - Interferometry Systems and Technology Phone: 818 354 2903 [EMAIL PROTECTED]
Re: [Cooker] gdm and kde
Do you have a link to the exact problem? Like at xfree86.org? or the kde guys? I can't seem to find a reference, but the solution works! :) Brad Felmey wrote: > On Thu, 2002-08-22 at 18:55, Pixel wrote: > >>Brad Felmey <[EMAIL PROTECTED]> writes: >> >> >>>On Thu, 2002-08-22 at 18:36, Brad Felmey wrote: >>> >>>>On Thu, 2002-08-22 at 18:31, Vincent Danen wrote: >>>> >>>> >>>>>Has anyone noticed any problems with gdm and KDE? I've got on my >>>>>dual-head machine here (two different cards), a problem when I load >>>>>KDE using gdm as my display manager... I get the desktop, but the >>>>>panel isnt' all there (no icons except for the screenlock and logout), >>>>>no menu. And screens sit strange; the splashscreen doesn't display, >>>>>and it doesn't disappear either (blank grey box that sits in the >>>>>foreground... moving windows around move *under* it). >>>>> >>>>>Using kdm, KDE seems to work fine, except kcontrol doesn't want to >>>>>launch for me (synched with cooker about 30 minutes ago). >>>> >>>[...] >>> >>>>Horrible XF 4.2 Xinerama bug. Kill off xfs and manually state the font >>>>paths in your /etc/X11/XF86Config-4, or turn off Xinerama. Your choice. >>> >>>This should be included in DrakX installer, methinks, at least until the >>>XF guys fix it. :) >> >>I'd rather have it fixed! Does it really comes from the use of xfs >>together with Xinerama? or even xfs + xineram + kde ? > > > It's xfs and xinerama in 4.2. Screws up in both GNOME and KDE. I can > duplicate it 100% on Matrox G450 cards. Drove me crazy until I ran > across the solution in one of the Debian mailing lists. This is not > Mandrake-specific - I've seen it in SuSE 8.0 and Debian Sid as well. > Solution is to not use xfs if Xinerama is turned on. Turn Xinerama off, > and there's no problem. This did/does not exist in XF 4.1, just 4.2. -- Bryan Whitehead SysAdmin - JPL - Interferometry Systems and Technology Phone: 818 354 2903 [EMAIL PROTECTED]
Re: [Cooker] RC2: Matrox Dualhead G450 crashes X11
Pixel wrote: > Reinhard Katzmann <[EMAIL PROTECTED]> writes: > > >>Ok I managed to get my DH card run with your BusID and the >>two "0 0" on the ServerLayout section (diff included). > > > I've add BusID's, but is the "0 0" really needed? It isn't here... Not in mine... just the BusID is needed for both heads. -- Bryan Whitehead SysAdmin - JPL - Interferometry Systems and Technology Phone: 818 354 2903 [EMAIL PROTECTED]
Re: [Cooker] beta1-rc2 mozilla 1.1 problems no solution.
> did you read my previous email ?? > > I've already told you our mozilla is build WITH Xinerama support ! > Check line 436 of mozilla specfile ... your right i missed it. > Only the gcc3.2 compiled library which are only used by OpenOffice.org > are not compiled with Xinerama and this is not an issue.. I expanded the beta and the lockups are having for everyone. Some the lockup only last a min, other have to kill off mozilla (or galeon). I dunno what else to report. But mozilla 1.1 is broken (for us anyway). Any suggestions? -- Bryan Whitehead SysAdmin - JPL - Interferometry Systems and Technology Phone: 818 354 2903 [EMAIL PROTECTED]
Re: [Cooker] Printing in MDK9.0 RC*
Eyal Ben-David wrote: > On Wednesday 11 September 2002 02:19, Till Kamppeter wrote: > >>LPRng is removed because CUPS contains all functionality to communicate >>with LPD/LPRng. To access a printer on a remote LPD/LPRng server choose >>"Remote LPD" in printerdrake. You need to know the server's host name or >>IP and the name of the print queue on the server. To make the local CUPS >>queues available for remote LPD/LPRng clients, activate the cups-lpd >>mini daemon by "chkconfig --add cups-lpd; service xinetd restart" or by >>using the "Services" section in the Mandrake Control Center. For the LPD >>clients your local CUPS queues have the same names under LPD as under >>CUPS. The clients can send PostScript jobs which get filtered by the >>CUPS server. >> >>Till >> > > > Hello Till, > > First let me thank you for the wonderful work at www.linuxprinting.org and > Mandrake printing software! > > Technically you are right. CUPS can do do whatever LPRng does. > but I think this is not a technical issue at all. Mandrake Linux has several > Window managers that do similar things, two versions of X11 (4.x and 3.x), > many kernels, many media players, office suites, graphic viewers etc etc. > > The point is choice and ML is doing right with all the packages mentioned > above. I think that the same argument must be applied to printing software. > You can say that only CUPS and PDQ are supported but you should give (IMO) > other people the option to select another package. Exactly! We have chosen not to use CUPS because it's far different from standard unix printing systems. We have a team of SA's and it's nice to not have too have a "CUPS guy" when stuff doesn't work. lpd and LPRng may not have an integrated webserver and magic integration with some very cool but mostly unused internet printing standard. But LPRng is a very nightweight printing system, it's secure, not undergoing tons of changes like CUPS is, and it just plain works. It's not that big of a deal, but it still is annoying. We have how many text editors? Too many to bother counting? Printing systems? Two. I feel like a vi user that just had vi removed... after all... can't emacs do all the same stuff? -- Bryan Whitehead SysAdmin - JPL - Interferometry Systems and Technology Phone: 818 354 2903 [EMAIL PROTECTED]
Re: [Cooker] beta1-rc2 mozilla 1.1 problems
Vincent Danen wrote:
>
> On Monday, September 9, 2002, at 04:00 PM, Bryan Whitehead wrote:
>
>> I think i traced down the problems I'm having with mozilla 1.1
>> --enable-xinerama is not being passed to configure at all...
>>
>> Mozilla locks up after about 30mins of use...
>>
>> also galeon locks up when starting up.
>>
>> I've trying blowing away .mozilla and .galeon and logging out (and
>> killing all my processes for good measure) and the problem is still
>> there. :(
>>
>> mozilla is very stable on a non-dual head box tho...
>
>
> I'm not sure this would have anything do with xinerama. I use Galeon
> only for my browser, and have multiple tabs and multiple windows open
> for hours (or days) at a time. Running pure cooker here with dual-head
> and xinerama enabled on GNOME2. I can say that Galeon has not crashed
> on me for quite a while, and I even used Mozilla yesterday for some
> testing and it was open for a few hours, and it didn't even blink.
>
> --
> MandrakeSoft Security; http://www.mandrakesecure.net/
> "lynx - source http://linsec.ca/vdanen.asc | gpg --import"
> {FE6F2AFD: 88D8 0D23 8D4B 3407 5BD7 66F9 2043 D0E5 FE6F 2AFD}
>
you can have dual head without having Xinnerama on. Also, your card
might have equal features on both heads. (For example on a Matrox g550
one one head has DRI support Xinerama communicates this to
applications that are Xinerama aware so they will work.) You might just
be lucky with your hardware. But I assure you, all dual heads here don't
work with current mozilla because --enable-xinerama is not on.
--
Bryan Whitehead
SysAdmin - JPL - Interferometry Systems and Technology
Phone: 818 354 2903
[EMAIL PROTECTED]
Re: [Cooker] beta1-rc2 mozilla 1.1 problems SOLUTION!
Frederic Crozat wrote: > Le Mon, 09 Sep 2002 15:00:35 +0000, Bryan Whitehead a écrit : > > >>I think i traced down the problems I'm having with mozilla 1.1 >>--enable-xinerama is not being passed to configure at all... > > > Really ? I don't know where you found that.. Mozilla IS compiled with > xinerama support.. > > Maybe it is a bug in your video card driver.. I rebuilt mozilla with --enable-xinerama and now mozilla is rock solid!!! PLEASE APPLY!!! Without this option mozilla crashes on all dual head machines we have!!! Here is a diff of the mozilla.spec: 53c53 < Release: 8mdk --- > Release: 9mdk 392c392 < --enable-old-abi-compat-wrappers --- > --enable-old-abi-compat-wrappers --enable-xinerama 431c431 < --enable-optimize="$OPT_FLAGS" \ --- > --enable-optimize="$OPT_FLAGS" --enable-xinerama \ -- Bryan Whitehead SysAdmin - JPL - Interferometry Systems and Technology Phone: 818 354 2903 [EMAIL PROTECTED]
Re: [Cooker] beta1-rc2 mozilla 1.1 problems
Frederic Crozat wrote: > Le Mon, 09 Sep 2002 15:00:35 +0000, Bryan Whitehead a écrit : > > >>I think i traced down the problems I'm having with mozilla 1.1 >>--enable-xinerama is not being passed to configure at all... > > > Really ? I don't know where you found that.. Mozilla IS compiled with > xinerama support.. > > Maybe it is a bug in your video card driver.. The spec file has no mention of --enable-xinerama being passed to configure. Galeon and Mozilla are completly unusable on any of our machines. We've tried removing the .mozilla and .galeon... But they will just lock up either when starting up, or within a couple of mins... (about 10). I'm at a loss of what's going on then. Everything else seems to work fine... using mozilla 1.0 from a different machine (via ssh) has no problems. Are any of you using the mail component? The mozilla mail client is pretty standard here. (as a side not the mozilla mail component will not lock up till an "heavy" html loaded email is displayed) Any idea's ? -- Bryan Whitehead SysAdmin - JPL - Interferometry Systems and Technology Phone: 818 354 2903 [EMAIL PROTECTED]
[Cooker] 9.0 RC2 drakx error (still)
(this is a dual-head, auto X config with kickstart). Ok, basically everything works except there is no "BusID" in the devices section Both devices need to have the busid listed in order for X to work. the -broke is the kickstart config the others are when I put BusID in the device section. I'm running a Matrox g550. http://huey.jpl.nasa.gov/~driver/ for the files (too big to attach them all I guess) -- Bryan Whitehead SysAdmin - JPL - Interferometry Systems and Technology Phone: 818 354 2903 [EMAIL PROTECTED]
[Cooker] beta1-rc2 mozilla 1.1 problems
I think i traced down the problems I'm having with mozilla 1.1 --enable-xinerama is not being passed to configure at all... Mozilla locks up after about 30mins of use... also galeon locks up when starting up. I've trying blowing away .mozilla and .galeon and logging out (and killing all my processes for good measure) and the problem is still there. :( mozilla is very stable on a non-dual head box tho... -- Bryan Whitehead SysAdmin - JPL - Interferometry Systems and Technology Phone: 818 354 2903 [EMAIL PROTECTED]
[Cooker] Printing in MDK9.0 RC*
What happened to the LPRng or a plain 'ol lpd? Is cups the one and only print option on MDK9.0 ? (Keep in mind I'm kickstarting and have not run the GUI installer since beta1) -- Bryan Whitehead SysAdmin - JPL - Interferometry Systems and Technology Phone: 818 354 2903 [EMAIL PROTECTED]
Re: [Cooker] RC1 kickstart error with Dual-Head.
>
>
>>>
>>>
>>>>After kickstarting X will not start.
>>>>
>>>>
>
>what exactly do you mean? even if you don't have Xinerama, it should
>start. Or is it the missing BusID that makes it go havoc?
>
It doesn't. It will just crash with signal 11... no explination. When
the BusID is listed for both heads (they are the same on my box) the
signal 11 goes away, but it crashes with an error becaise dri is active
on the second head. (even when xinerama is off). So turning dri off
makes X work
However, since it's not xinerama kdm looks like crap at the login screen
with the second monitor black. when you login 2 desktops are created
under KDE because it sees 2 screens, not a Xinerama... (that might be on
purpose). Gnome has problems placing windows in the middle of the
"desktop" so every window must be manually moved to keep them on one
screen. :(
Is it too much to have a "if dual-head use xinerama" ?
>In auto_install, i can't default on Xinerama since it means no DRI and
>DRI is more often used than Xinerama.
>
>
DRI will not work with Dual Head anyway. so no loss. at least with
matrox cards (actually you have have dri work for only the first
head, but not the second, but I don't know how todo that... :-P )
>I'm adding an option in auto_install to prefer Xinerama:
>
>$o = {
> X => { Xinerama => 1 },
>};
>
>
>
>>BusID must be set for both heads (they are the same value on my machine).
>>
>>
>
>this one is weird. In tests i've done it was not needed. Fredl, WYT?
>
If I don't have BusID set for both heads it will core dump with signal 11.
I can provide core files if needed... ;)
--
Bryan Whitehead
SysAdmin - JPL - Interferometry Systems and Technology
Phone: 818 354 2903
[EMAIL PROTECTED]
[Cooker] RC1 kickstart error with Dual-Head.
After kickstarting X will not start. XF86Config-4.old is the kickstart generated config file that does not work. XF86Config-4 is the config file I pulled from a mandrake 8.1 box. Please fix. :) -- Bryan Whitehead SysAdmin - JPL - Interferometry Systems and Technology Phone: 818 354 2903 [EMAIL PROTECTED] # File generated by XFdrake. # ** # Refer to the XF86Config man page for details about the format of # this file. # ** Section "Files" # Multiple FontPath entries are allowed (they are concatenated together) # By default, Mandrake 6.0 and later now use a font server independent of # the X server to render fonts. FontPath "unix/:-1" EndSection Section "ServerFlags" #DontZap # disable (server abort) AllowMouseOpenFail # allows the server to start up even if the mouse doesn't work #DontZoom # disable / (resolution switching) EndSection Section "Module" Load "dbe" # Double-Buffering Extension Load "v4l" # Video for Linux Load "extmod" Load "type1" Load "freetype" Load "glx" # 3D layer Load "dri" # direct rendering EndSection Section "DRI" Mode 0666 EndSection Section "InputDevice" Identifier "Keyboard1" Driver "Keyboard" Option "XkbModel" "pc105" Option "XkbLayout" "us" Option "XkbOptions" "" EndSection Section "InputDevice" Identifier "Mouse1" Driver "mouse" Option "Protocol" "PS/2" Option "Device" "/dev/psaux" Option "Emulate3Buttons" Option "Emulate3Timeout" "50" EndSection Section "Monitor" Identifier "monitor1" VendorName "Plug'n Play" ModelName "Dell P1110" HorizSync 30.0-121.0 VertRefresh 48.0-160.0 # Sony Vaio C1(X,XS,VE,VN)? # 1024x480 @ 85.6 Hz, 48 kHz hsync ModeLine "1024x480"65.00 1024 1032 1176 1344 480 488 494 563 -hsync -vsync # TV fullscreen mode or DVD fullscreen output. # 768x576 @ 79 Hz, 50 kHz hsync ModeLine "768x576" 50.00 768 832 846 1000 576 590 595 630 # 768x576 @ 100 Hz, 61.6 kHz hsync ModeLine "768x576" 63.07 768 800 960 1024 576 578 590 616 EndSection Section "Monitor" Identifier "monitor2" # Sony Vaio C1(X,XS,VE,VN)? # 1024x480 @ 85.6 Hz, 48 kHz hsync ModeLine "1024x480"65.00 1024 1032 1176 1344 480 488 494 563 -hsync -vsync # TV fullscreen mode or DVD fullscreen output. # 768x576 @ 79 Hz, 50 kHz hsync ModeLine "768x576" 50.00 768 832 846 1000 576 590 595 630 # 768x576 @ 100 Hz, 61.6 kHz hsync ModeLine "768x576" 63.07 768 800 960 1024 576 578 590 616 EndSection Section "Device" Identifier "device1" VendorName "Matrox" BoardName "Matrox Millennium G550 DualHead" Driver "mga" VideoRam 32768 Screen 0 Option "DPMS" Option "AGPMode" "true" EndSection Section "Device" Identifier "device2" VendorName "Matrox" BoardName "Matrox Millennium G550 DualHead" Driver "mga" Screen 1 Option "DPMS" EndSection Section "Screen" Identifier "screen1" Device "device1" Monitor "monitor1" DefaultColorDepth 24 Subsection "Display" Depth 8 Modes "1600x1200" "1400x1050" "1280x1024" "1280x960" "1152x864" "1024x768" "800x600" "640x480" EndSubsection Subsection "Display" Depth 15 Modes "1600x1200" "1400x1050" "1280x1024" "1280x960" "1152x864" "1024x768" "800x600" "640x480" EndSubsection Subsection "Display" Depth 16 Modes "1600x1200" "1400x1050" "1280x1024" "1280x960" "1152x864" "1024x768" "800x600" "640x480" EndSubsection Subsection "Display" Depth 24 Modes "1600x1200" "1400x1050" "1280x1024" "1280x960" "1152x864" "1024x768" "800x600" "640x480" EndSubsection EndSection Section "Screen" Identifier "scr
Re: [Cooker] 9.0 RC1 soundcard and scanner not installed, wrong HOSTNAME,errors during boot
David Walser wrote: > --- Stefan Hußfeldt <[EMAIL PROTECTED]> > wrote: > >>Hello. >> >>Installation (Mandrake Linux Cooker-i586 20020903 >>9:18) via network (FTP, >>DHCP): >> >>- HOSTNAME is set to localhost > > > I ran into this too. In case you don't know what he's > talking about, if you choose DHCP for network settings > in network.img, it asks for an optional hostname and > domain, but it doesn't keep those settings for later. > > __ > Do You Yahoo!? > Yahoo! Finance - Get real-time stock quotes > http://finance.yahoo.com > Also, if the hostname is set via DHCP mandrake ignores. This has been a problem since 8.2... -- Bryan Whitehead SysAdmin - JPL - Interferometry Systems and Technology Phone: 818 354 2903 [EMAIL PROTECTED]
Re: [Cooker] thoughts on mp3 issue
[EMAIL PROTECTED] wrote: > That's not the point here. Besides, you got a spare $50,000 for > Mandrake?? As it is they are AFAIK laying off developers and (perhaps) > rushing releases a bit just to stay afloat. I am not ripping ML, just > trying to be honest. I do NOT want to see ML go under, THAT would be a > sad day. So if it comes down to a PLF package instead of ML paying for > MP3's, I'll take it. > > Regards, > If the cost was split between most linux distro's, as well as community support... It souldn't be that much. Even if it was only split 5 ways Mandrake would need only $10k. And that's chump change next to the salary of one employee. I was just throwing it up as a "maybe this is an option". A lawsuit from the owners of the mp3 licence is far worse than just paying the entire $50k It plain sucks no matter how you look at it. -- Bryan Whitehead SysAdmin - JPL - Interferometry Systems and Technology Phone: 818 354 2903 [EMAIL PROTECTED]
Re: [Cooker] thoughts on mp3 issue
Guillaume Cottenceau wrote: > [EMAIL PROTECTED] writes: > > >>S, is XMMS going to play MP3's or not in 9.0??? > > > dunno yet. we're studying the issue with lawyers, and contacting > thomson & rh to get more info on the subject. > Why doesn't Mandrake, RedHat, and others simply pay $50,000 on behalf of the XMMS team. Then they will have an unlimited license for decoding .mp3's. Mandrake/Redhat/others simply distribute XMMS Or am I missing something? -- Bryan Whitehead SysAdmin - JPL - Interferometry Systems and Technology Phone: 818 354 2903 [EMAIL PROTECTED]
Re: [Cooker] NFS Installation MDK 9 B4
Pablo Pita Leira wrote: > I have installed 9 Beta 4 in a laptop. I did a NFS installation. > > I found these problems : > > 1.- I entered the hostname and domain name to use the network, but after doing > the installation, the hostname is setup as localhost. The domain name was > correct. I'll second this error. I've reported it before. I think it's an issue with DHCP... But I could be wrong. > 2.- I use a PCMCIA ethernet card based on a 8139 chip. The 8139too driver > works with the card (it was used for the installation). After the > installation, the pcmcia card manager loads the 8139cp driver which fails. > -- Bryan Whitehead SysAdmin - JPL - Interferometry Systems and Technology Phone: 818 354 2903 [EMAIL PROTECTED]
Re: [Cooker] RPM volnorm plugin for xmms
Levi Ramsey wrote: > On Wed Aug 21 11:32 +0200, Buchan Milne wrote: > >>Bryan Whitehead wrote: >> >>>Should be included. >>> >>>some of us live in cubicles at work, and sometimes mp3's have screwed up >>>volumes. >>> >>>This plugin keeps all audio comming out of xmms at the same volume. >>> >>>src filename: volnorm-0.8.1-2mdk.src.rpm >>>i586 filename: volnorm-0.8.1-2mdk.i586.rpm >>> >>>located: ftp://ftp.linux-mandrake.com/incoming/ >>> >> >>Have you tried xmms-normalize (in contrib AFAIK)? > > > I've had very good luck with xmms-compress (from > http://trikuare.cx/~magenta/projects/xmms-compress.html ). > I couldn't find a mandrake packgae, but I think this one is by far the best :) -- Bryan Whitehead SysAdmin - JPL - Interferometry Systems and Technology Phone: 818 354 2903 [EMAIL PROTECTED]
Re: [Cooker] RPM volnorm plugin for xmms
Buchan Milne wrote: > Bryan Whitehead wrote: > >> Should be included. >> >> some of us live in cubicles at work, and sometimes mp3's have screwed >> up volumes. >> >> This plugin keeps all audio comming out of xmms at the same volume. >> >> src filename: volnorm-0.8.1-2mdk.src.rpm >> i586 filename: volnorm-0.8.1-2mdk.i586.rpm >> >> located: ftp://ftp.linux-mandrake.com/incoming/ >> > Have you tried xmms-normalize (in contrib AFAIK)? > I guess it was already packaged! The name is wrong tho.. "normalize" is a different peice of software. volnorm is the correct name. http://volnorm.sourceforge.net/ "normalize" is a simular project, based off of his project. http://volnorm.sourceforge.net/others.html -- Bryan Whitehead SysAdmin - JPL - Interferometry Systems and Technology Phone: 818 354 2903 [EMAIL PROTECTED]
[Cooker] RPM volnorm plugin for xmms
Should be included. some of us live in cubicles at work, and sometimes mp3's have screwed up volumes. This plugin keeps all audio comming out of xmms at the same volume. src filename: volnorm-0.8.1-2mdk.src.rpm i586 filename: volnorm-0.8.1-2mdk.i586.rpm located: ftp://ftp.linux-mandrake.com/incoming/ -- Bryan Whitehead SysAdmin - JPL - Interferometry Systems and Technology Phone: 818 354 2903 [EMAIL PROTECTED]
Re: [Cooker] M9 Beta3 Mozilla bug report
>>yes yes... I'm aware of that, but Mozilla.org is not exactly "on time" >>with there releases. Hopefully they have changed, but they have no track >>record of releasing on time. 1.0 was late, but good. I suspect the same >>with 1.1, late but good. The problem is 1.1 isn't ready yet, and the >>"ideal release" date has come and gone. If they are "on time" just like >>like thier other releases (1.1 alpha nad beta) we can expect thier 1.1 >>final in 3 days. That's fine., but leavs little time for testing for >>mandrake and us. > > > If you were reading mozillazine and mozilla newsgroup, you would know that > release candidate of Mozilla 1.1 are out and 1.1 should ship probably next > monday or tuesday.. Like I said, final should be out in 3 days > Guys, you are incredible : for Mdk 8.2, I was bashed because I didn't > include Moz 0.9.8 (and I was right, there was a lot of regressions > compared to Moz 0.9.7 we shipped).. > > And now that we will ship Moz 1.1, we are bashed because it is too recent > (even if it has passed both alpha and beta stage).. You are never happy and > I'm becoming a little upset by this kind of behaviour.. You didn't release with 0.9.8 because you didn't have enough time to test it... I don't want the same thing to happen with mozilla 1.1 where we end up with mozilla-crap-o-rama-beta-rc because they took too long and you don't have time to test the final. You didn't ship with 0.9.8 becasue it was to close to a release... And I'm fine with that. I understand. I'm letting you know that the current mozilla (whatever version) is crap, and crashes all the time. And I also understand you need time to test the RC or final... > Did you see any Release Candidate of Mandrake Linux recently ? There won't > be "final" Mdk 9.0 until at least some RC are out.. My main concern is having a mozilla - whatever the version is - stable. And I'm on this list to test, for myself, and you. -- Bryan Whitehead SysAdmin - JPL - Interferometry Systems and Technology Phone: 818 354 2903 [EMAIL PROTECTED]
Re: [Cooker] M9 Beta 2 Bug2 - No Dual Head support
Pixel wrote:
> Bryan Whitehead <[EMAIL PROTECTED]> writes:
>
>
>>Pixel wrote:
>>
>>>Bryan Whitehead <[EMAIL PROTECTED]> writes:
>>>[...]
>>>
>>>
>>>>I have a Matrox g550 and Beta 3 still does not work for multihead. I have dual
>>>>21" Trinitron Monitors hooked up (same model/brand).
>>>
>>>what does "lspcidrake -v" gives ?
>>
>>[...]
>>Card:Matrox Millennium G550: Matrox|MGA G550 AGP [DISPLAY_VGA] (vendor:102b
>>device:2527 subv:102b subd:0f84)
>>[...]
>>Keep in mind I'm kickstarting from an auto_inst.cfg file. If you need that
>>fine or any other info let me know.
>
>
> - does XFdrake configures multi-head properly?
If I run XFdrake after the machine is kickstarted I cannot configure
Dual-Head. On the same machine Mandrake 8.1 (useing XFdrake) would
configure the dual head no problem.
> - in auto_install it should "Configure all heads independently", isn't
> that so?
Most of our machines are single head. Only 5 dual-head. auto_inst.cfg
has this for X:
'X' => {
'resolution_wanted' => '1600x1200',
'default_depth' => '24'
},
Note: we only started using dual head in the last several months on
linux (we have many sun boxes with dual head - but solaris is another
story). So Mandrake 9.0 is the first time we've kickstarted a dual head
machine.
--
Bryan Whitehead
SysAdmin - JPL - Interferometry Systems and Technology
Phone: 818 354 2903
[EMAIL PROTECTED]
Re: [Cooker] M9 Beta 2 Bug2 - No Dual Head support
Pixel wrote: > Bryan Whitehead <[EMAIL PROTECTED]> writes: > > [...] > > >>I have a Matrox g550 and Beta 3 still does not work for multihead. I have dual >>21" Trinitron Monitors hooked up (same model/brand). > > > what does "lspcidrake -v" gives ? > agpgart : Intel Corporation|82860 860 (Wombat) Chipset Host Bridge (MCH) [BRIDGE_HOST] (vendor:8086 device:2531 subv:1028 subd:00d8) unknown : Intel Corporation|82850 850 (Tehama) Chipset AGP Bridge [BRIDGE_PCI] (vendor:8086 device:2532) unknown : Intel Corporation|82860 860 (Wombat) Chipset AGP Bridge [BRIDGE_PCI] (vendor:8086 device:2533) unknown : Intel Corporation|82820 815e (Camino 2) Chipset PCI [BRIDGE_PCI] (vendor:8086 device:244e) i810-tco: Intel Corporation|82820 815e (Camino 2) Chipset ISA Bridge (ICH2) [BRIDGE_ISA] (vendor:8086 device:2440) unknown : Intel Corporation|82820 815e (Camino 2) Chipset IDE U100 [STORAGE_IDE] (vendor:8086 device:244b subv:1028 subd:00d8) usb-uhci: Intel Corporation|82820 815e (Camino 2) Chipset USB (Hub A) [SERIAL_USB] (vendor:8086 device:2442 subv:1028 subd:00d8) unknown : Intel Corporation|82820 815e (Camino 2) Chipset SMBus [SERIAL_SMBUS] (vendor:8086 device:2443 subv:1028 subd:00d8) usb-uhci: Intel Corporation|82820 815e (Camino 2) Chipset USB (Hub B) [SERIAL_USB] (vendor:8086 device:2444 subv:1028 subd:00d8) Card:Matrox Millennium G550: Matrox|MGA G550 AGP [DISPLAY_VGA] (vendor:102b device:2527 subv:102b subd:0f84) unknown : Intel Corporation|82806AA PCI64 Hub PCI Bridge [BRIDGE_PCI] (vendor:8086 device:1360) unknown : Intel Corporation|82806AA PCI64 Hub Advanced Programmable Interrupt Controller [SYSTEM_PIC] (vendor:8086 device:1161) aic7xxx : Adaptec|7892P [STORAGE_SCSI] (vendor:9005 device:008f subv:1028 subd:00d8) 3c59x : 3Com Corporation|3c905C-TX [Fast Etherlink] [NETWORK_ETHERNET] (vendor:10b7 device:9200 subv:1028 subd:00d8) ohci1394: Texas Instruments|TSB12LV26 OHCI-Lynx PCI IEEE 1394 Host Controller [SERIAL_FIREWIRE] (vendor:104c device:8020 subv:1028 subd:00d8) snd-emu10k1 : Creative Labs|SB Live! (audio) [MULTIMEDIA_AUDIO] (vendor:1102 device:0002 subv:1102 subd:8027) emu10k1-gp : Creative Labs|SB Live! (joystick) [INPUT_OTHER] (vendor:1102 device:7002 subv:1102 subd:0020) unknown : Virtual|Hub [] (vendor: device:) unknown : Virtual|Hub [] (vendor: device:) Keep in mind I'm kickstarting from an auto_inst.cfg file. If you need that fine or any other info let me know. -- Bryan Whitehead SysAdmin - JPL - Interferometry Systems and Technology Phone: 818 354 2903 [EMAIL PROTECTED]
Re: [Cooker] M9 Beta3 Mozilla bug report
Frederic Crozat wrote: > Le Fri, 16 Aug 2002 10:56:19 +0000, Bryan Whitehead a écrit : > > >>I'm not sure how wise it is that your all using Mozilla 1.1Beta... We >>expanded our beta testbed to actual users and the biggest complaint is >>Mozilla locks up about 5 times an hour while just using the mail >>component. >> >>Unless Mozilla 1.1 is going to reach final maybe it would be better to >>fall back to Mozilla 1.0 ? > > > Of course, if you had read the changelog entry, you had seen we are using > the almost final version of Mozilla 1.1 (ie branch point from last week) > and not the 1.1beta.. Moreover, Mozilla 1.1 should be released next week > (and changes are minimal with the current branch version in > beta3/cooker).. yes yes... I'm aware of that, but Mozilla.org is not exactly "on time" with there releases. Hopefully they have changed, but they have no track record of releasing on time. 1.0 was late, but good. I suspect the same with 1.1, late but good. The problem is 1.1 isn't ready yet, and the "ideal release" date has come and gone. If they are "on time" just like like thier other releases (1.1 alpha nad beta) we can expect thier 1.1 final in 3 days. That's fine., but leavs little time for testing for mandrake and us. It really wouldn't worry me at all except the last Mandrake releases have been "final" without any warning and I don't want to come in monday morning with a "Mandrake 9.0 Fianl Released" (as happened with Mandrake 8.2) with a bunch of stuff broke - like an "almost final" version of mozilla. So if you know 9.0 Final is going to wait for the mozilla folk to cut final... great. Otherwise the beta/cvs-snapshot isn't cutting it. -- Bryan Whitehead SysAdmin - JPL - Interferometry Systems and Technology Phone: 818 354 2903 [EMAIL PROTECTED]
Re: [Cooker] M9 Beta 2 Bug2 - No Dual Head support
Pixel wrote: > "Adam Flinton" <[EMAIL PROTECTED]> writes: > > >>I have a Matrox G550 & there is no option either in the install or >>anywhere else (that I can see) to configure Dual Head support (independant >>or Xinerama). >> >>It is totally missing from the install & if there is a GUI to config Dual >>head in M9B2 could someone please point me to it? > > > It should be fixed in beta3 (hardware detection pb) > > No way to force it :-/ (i won't change this now, too risky) > I have a Matrox g550 and Beta 3 still does not work for multihead. I have dual 21" Trinitron Monitors hooked up (same model/brand). I had to copy my XF86Config-4 from a different workstation (that has the same configuration) to get X working. DrakX simply refuses todo dual head on the MatroxG550... Weird because it worked just fine in 8.1. -- Bryan Whitehead SysAdmin - JPL - Interferometry Systems and Technology Phone: 818 354 2903 [EMAIL PROTECTED]
Re: [Cooker] Beta 3 (ouch!): depmod error
Richard Tango-Lowy wrote: > depmod -a gives > > depmod: *** Unresolved symbols in > /lib/modules/2.4.19-3mdk/kernel/drivers/media/video/saa7134.o.gz > Rich > > ars Cognita The Art of Knowledge > - > Richard Tango-Lowy > [EMAIL PROTECTED] > 603 424-6555 > > > I'll second that error. ;) -- Bryan Whitehead SysAdmin - JPL - Interferometry Systems and Technology Phone: 818 354 2903 [EMAIL PROTECTED]
[Cooker] mdk9.0 Beta3 X error init level 5
X does not automatically start in init level 5 after kickstart. I looked at /etc/inittab and some lines are missing (mdk8.1 has these lines as does some random cooker from 1-2 weeks ago): # Run xdm in runlevel 5 # xdm is now a separate service x:5:respawn:/etc/X11/prefdm -nodaemon So now X will not start up in init level 5. Is this just a kickstart problem? Does some drakx script modify /etc/inittab? Or is gdm/kdm/xdm being started differently now? -- Bryan Whitehead SysAdmin - JPL - Interferometry Systems and Technology Phone: 818 354 2903 [EMAIL PROTECTED]
Re: [Cooker] Beta2 of 9.0 kickstart error
Bryan Whitehead wrote: > I've just kickstarted our test machines with the new mdk9.0 Beta 2 and > now # of CPU detection has failed. Normally Mandrake would correctly see > that we have more than 1 CPU and would install the SMP kernel. Beta 2 > does not do this. :( > > i found the problem as we moved from Beta1 to Beta2 we used the same auto_inst.cfg file. In order for mandrake to correctly figure out and configure the correct kernel all kernel "options" must be installed. That means that in the package list kernel, kernel-smp, and kernel-enterprise must be in the list. The problem is with the "new" mandrake way of naming kernel packages. "kernel" is now "kernel--?mdk" and kernel-smp is "kernel-smp--?mdk" etc etc. So everytime a new Beta comes our or I update cooker I have to go in and hand edit the package list to be correct Anyhow... Maybe there is a better way to make the # of CPU code / how much memory work? -- Bryan Whitehead SysAdmin - JPL - Interferometry Systems and Technology Phone: 818 354 2903 [EMAIL PROTECTED]
Re: [Cooker] Beta2 of 9.0 kickstart error
Randy wrote: > Bryan Whitehead wrote: > >> Pixel wrote: >> >>> Bryan Whitehead <[EMAIL PROTECTED]> writes: >>> >>> >>>> I've just kickstarted our test machines with the new mdk9.0 Beta 2 >>>> and now # >>>> of CPU detection has failed. Normally Mandrake would correctly see >>>> that we >>>> have more than 1 CPU and would install the SMP kernel. Beta 2 does >>>> not do >>>> this. :( >>> >>> >>> i really don't know how to fix those pbs. I may find the time to sync >>> the SMP detection with redhat's ... But I'm not sure it will fix it >>> since it's really kernel dependent :'-( >> >> >> It's worked in mdk8.1, 8.2, 9.0beta1, and most of cooker. (till beta2 >> was released). >> >> So I know it can be done... was the kernel changed so much? >> >> Also when memory is over the 900MB it doesn't install the enterprize >> kernel... > > > Mandrake correctly detected my two CPUs in 8.2, 9.0b1, 9.0b2 perfectly, > and when I upgraded my memory to 1GB the 9.0b2 installer picked that > fact up and installed the enterprise kernel, no problem. Note that > I am not using kickstart. > > Rand. > Tiger MPX w/Athlon MPs > > I'm testing on Dual Pentium III Machines (800MHZ). Maybe your dual athlon is easier? hmmm... I'm syncing with latest cooker now for another test. maybe I should test on my own machine (Dual P4-XEON 1.7Ghz) perhaps just Pentium III's have a problem? -- Bryan Whitehead SysAdmin - JPL - Interferometry Systems and Technology Phone: 818 354 2903 [EMAIL PROTECTED]
Re: [Cooker] Beta2 of 9.0 kickstart error
Pixel wrote: > Bryan Whitehead <[EMAIL PROTECTED]> writes: > > >>I've just kickstarted our test machines with the new mdk9.0 Beta 2 and now # >>of CPU detection has failed. Normally Mandrake would correctly see that we >>have more than 1 CPU and would install the SMP kernel. Beta 2 does not do >>this. :( > > > i really don't know how to fix those pbs. I may find the time to sync > the SMP detection with redhat's ... But I'm not sure it will fix it > since it's really kernel dependent :'-( > It's worked in mdk8.1, 8.2, 9.0beta1, and most of cooker. (till beta2 was released). So I know it can be done... was the kernel changed so much? Also when memory is over the 900MB it doesn't install the enterprize kernel... -- Bryan Whitehead SysAdmin - JPL - Interferometry Systems and Technology Phone: 818 354 2903 [EMAIL PROTECTED]
[Cooker] mdk9.0 Beta1 bug
On all previous versions of Mandrake, from 7.0 to 8.1 when a machine was
given an IP from DHCP the hostname would be set correctly.
As of Mandrake 8.2, (and 9.0b1) when an IP is given over DHCP the
hostname will be "localhost". Even when it has a fully qualified domain
name.
This is very irritating. We kickstart workstations using DHCP then on
reboot run a series of scripts to configure the machines. (Basically a
post kickstart configuration). Most configureations (including a script
that changes the machine from a DHCP client to a static IP) break when
the hostname isn't set. We end up having to write ugle messing scripts
to find out hostname.
Example, On Mandrake 8.1 when using DHCP to get an IP:
[driver@mulan ~]$ hostname
mulan.jpl.nasa.gov
[driver@mulan ~]$
Machine using Mandrake 8.2 using DHCP to get an IP:
[driver@localhost ~]$ hostname
localhost.jpl.nasa.gov
[driver@localhost ~]$
Previously we'd need to do this for our scripts:
hostname=`hostname -a`
ip=`host $hostname | awk '{ print $4 }'`
Since Mandrake 8.2 things got messy:
ip=`ip addr show dev eth0 | grep inet | awk '{print $2}' | sed "s/\/.*//"`
hostname=`host $ip | awk '{print $5}' | sed "s/\..*//"`
The side effect of having the machine think it's localhost also causes
problems when configureing various services.
I guess my gripe is... I don't like things to be so messy... ;)
--
Bryan Whitehead
SysAdmin - JPL - Interferometry Systems and Technology
Phone: 818 354 2903
[EMAIL PROTECTED]
Re: [Cooker] newest dump/restore released yesterday
Buchan Milne wrote: > -BEGIN PGP SIGNED MESSAGE- > Hash: SHA1 > > Well, then not everyone has working dump/restore. Last time I checked, > the opinion of lkml was that dump/restore (except on XFS using > xfsdump/xfsrestore) WILL NOT WORK in 2.4. Ever. You'll just have very bad dumps. If the file is open in read-write mode the file might still be dirty in memory, so the dump will have a bad file. If you use ext3 all data is commited to disk right away, so your dumps will be "more clean". But yea, XFS is the only FS on linux with a true working dump/restore. > So that's why I am asking what the point is of updating non-xfs > dump/restore. > > Maybe I am wrong, but that's my reason for asking. Dump for ext2 is pretty crappy. on an active filesystem you will have files that are "backup up" that are just crap But chances are you don't have all your files open and being chanded anyway... So a dump on ext2 is crappy, but better than nothing. :) Read this thread on lkml, it started a big flamewar about backups: http://www.uwsg.iu.edu/hypermail/linux/kernel/0207.2/0020.html -- Bryan Whitehead SysAdmin - JPL - Interferometry Systems and Technology Phone: 818 354 2903 [EMAIL PROTECTED]
Re: [Cooker] mdk9.0 DVD-R/RW ??
Warly wrote: > FACORAT Fabrice <[EMAIL PROTECTED]> writes: > > >>Le Mercredi 31 Juillet 2002 11:42, Warly a écrit : >> >> >>>DVD+RW are a different media (however latest DVD burner are >>>able to burn DVD-RW and DVD+RW IIRC). It can be use as block >>>devices and directly handled by the kernel as any removable >>>devices. >> >>what does this mean ? >>does this mean that if I put a DVD+RW in a DVD+RW recorder I will be able to >>write data as if it was a HD/zip/floppy/ with a single cp/mv ? > > > by memory I think so (to be checked however) > The website I gave says that this will not work unless the patches are applied. As of Mandrake 8.0-8.2 this was true, we had to apply the patches ourselvs to get the DVD-RW to work. We could not just mount a disk and start writing to it. -- Bryan Whitehead SysAdmin - JPL - Interferometry Systems and Technology Phone: 818 354 2903 [EMAIL PROTECTED]
Re: [Cooker] Error with kickstart mdk9.0 beta
Pixel wrote: > Bryan Whitehead <[EMAIL PROTECTED]> writes: > > >>* warning: INTERNAL ERROR: >>common::internal_error() called from /usr/bin/perl-install/Xconfigurator.pm:224 >>Xconfigurator::multi_head_choose() called from > > > i've changed this quite a lot. Could you test with current cooker? > I can't get a goos sync with ftp.rpmfind.net or rpmfind.net that has a functioning cooker. I get (in part not full text) "Can't locate partition_table_raw.pm on @INC (@INC contains: " I searched for partition_table_raw.pm and it is not in the cooker tree from those 2 mirrors... is there a better mirror? or am I supposed todo something else? -- Bryan Whitehead SysAdmin - JPL - Interferometry Systems and Technology Phone: 818 354 2903 [EMAIL PROTECTED]
[Cooker] mdk9.0 network change/suggestion
When configureing laptops, typically we have 2 cards in the laptop. One
is a wired 10/100 connectio, and the other is a wireless card.
When a user is in thier office they use the wired card, wireless access
points are not available in offices/labs. But when they walk to an
conference room, or "special area's" on campus they'd like to have the
wireless card "just work".
Basically from the users perspective they should not care what interface
is working. Both cards should be pluged in at all times, and
internet/lan access should just work. Also, if the wired card is plugged
in, the default should be to use the wired connection and ignore the
wireless card. As soon as they unplug their wired connection the
wireless card should take over and everything should "fall back" without
any intervention from the user.
This is possible with a some simple changes to the ifup script. And will
work even better with some options turned on in the kernel.
In the ifup script "ip route add default via ${GATEWAY}" and "ip route
add default dev ${REALDEVICE}" need to be changed to: "ip route add
default via ${GATEWAY} equalize" and "ip route add default dev
${REALDEVICE} equalize". Basically we need the default rout to be able
to switch.
This is the change needed for static address. (it gets messy for dynamic)
Note, the primary interface (the wired card or eth0) can be set to
equalize and the other interface can be a regular (no equalize). This is
how we currently set up the laptops. since eth0 (wired) is the "first
default route" it will get all the traffic, but when it's link goes down
the wireless card will take over. Since the wireless card is not
equalized (it's dynamic and dhcp does not pass the equalize statement)
it will remain default once it switched over. So we tell users when they
get to thier office to remove the wireless pcmcia card can reinsert
after they pluyg thier laptops back in. (this is the same behavior that
windows 2000 has, once the secondary interface takes over it won't go
back till reboot - or removing and reinserting the card - so this isn't
a "big deal").
There is also some stuff in sysctl.conf that needs to be set.
net.ipv4.ip_forward = 1
net.ipv4.conf.default.rp_filter = 0
net.ipv4.route.gc_timeout = 10
the gc_timeout will insure that when the fallback does happen the route
cache does not try to timeout all over again with the route that is down.
If you guys know how to hook into a dhcp client to pass equalize to the
default route then having a laptop (or desktop) with mutiple default
routes will failover very nicely! :)
Currently I manually modify /ets/sysctl.conf and
/etc/sysconfig/network-scripts/ifup myself... but I think it's a great
feature to add to mandrake 9.0 as default
--
Bryan Whitehead
SysAdmin - JPL - Interferometry Systems and Technology
Phone: 818 354 2903
[EMAIL PROTECTED]
[Cooker] Error with kickstart mdk9.0 beta
Kickstart fails when using NVidia graphics cards. (Maybe it's all graphics cards but we only have Nvidia on out test machines) This is the error from the kickstart: * step `setupBootloader' finished * starting step `createBootdisk' * to put in modules scsi_hostadapter * step `createBootdisk' finished * starting step `configureX' * writing /mnt/etc/fstab * to put in modules scsi_hostadapter * selecting packages * examining packagesToInstall of medium disc 1 MandrakeLinux 9.0 beta1 (x86) * examining packagesToInstall of medium disc 3 Mandrake Linux 9.0 beta 1 (x86) * examining packagesToInstall of medium disc 2 Mandrake Linux 9.0 beta 1 (x86) * found 0 packages to install * all packages selected are already installed, nothing to do * unknown package `NVIDIA_GLX' * class_discard: class_discard::wait_message called at Xconfigurator/usr/bin/perl-install/Xconfigurator.pm1317 * class_discard: class_discard::ask_from_listf called at Xconfigurator/usr/bin/perl-install/Xconfigurator.pm222 * warning: INTERNAL ERROR: common::internal_error() called from /usr/bin/perl-install/Xconfigurator.pm:224 Xconfigurator::multi_head_choose() called from /usr/bin/perl-install/Xconfigurator.pm:395 Xconfigurator::card_config() called from /usr/bin/perl-install/Xconfigurator.pm:1319 Xconfigurator::main() called from /usr/bin/perl-install/install_steps.pm:915 install_steps::configureX() called from /usr/bin/perl-install/install2.pm:78 install2::installStepsCall() called from /usr/bin/perl-install/install2.pm:307 install2::configureX() called from /usr/bin/perl-install/install2.pm:566 (eval)() called from /usr/bin/perl-install/install2.pm:565 install2::main() called from /usr/bin/runinstall2:36 * class_discard: class_discard::DESTROY called at install2/usr/bin/perl-install/install2.pm570 Warning: String "include-menu-defs" didn't occur in example file /usr/X11R6/lib/X11/mwm//system.mwmrc-menu I can sent the entire ddebug.log if needed and auto_inst.cfg file if needed. Looks like a problem with multi_head_choose() not getting interactive input - which we do not want todo... -- Bryan Whitehead SysAdmin - JPL - Interferometry Systems and Technology Phone: 818 354 2903 [EMAIL PROTECTED]
[Cooker] Re: [expert] XFS and Linux-Mandrake
I can reproduce the problem constantly.. civileme has asked for scripts, but it will take me awhile to provide them as they are production scripts for our "jumpstart"... Seems to only be triggered on dual machines that are very fast. (2X 1.2-2.2Ghz p4 Xeons) (Basically all our new machines we get in). ext2 doesn't have the problem but XFS does... so I don't think it's a vfs issue... The problem "seems" to have gone away (as we havn't seen it yet) on Mandrake 8.2... On Wed, 2002-04-10 at 11:27, Guy Zelck wrote: > Bryan Whitehead wrote: > > >On Tue, 2002-04-02 at 04:04, Guy Zelck wrote: > > > >>You obviously haven't followed my lead : read the 'w > >>ich is better choice ext3 or ...' thread. A lot of detail is in there. > >>I can reproduce it any time by just doing un unclean shutdown or when I have to >when the system hangs. But maybe it has something to do with my SCSI card which is >not 100%. I'm getting closer. > >> > >>Guy. > >> > > > >I should have mentioned I read the thread. We have had problems on > >machines that shutdown cleanly Files end up with null's in them > >instead of data. I can reproduce the error on basically any of our > >machines (around 20 Dell machines with onboard scsi). The problem never > >comes up with other FS's. (well, at least ext2) On very heavily loaded > >machines we've had entire directories disappear. Running any of the xfs > >tools does not restore missing directories or fix files that are full of > >null characters. > > > >If anyone would like instructions on how we trigger the problem about > >30% of the time (for null characters), I can give detailed instructions > >and some scripts. It would be nice to have this resolved. We've had to > >switch back to ext2 since the problems were discovered. > > > Strange isn't it how important problems seem to be ignored sometimes. > I've got the same problems in md8.1 kernel 2.4.8. I don't need scripts > to provoke it, any unclean shutdown does it. > Although my system disk is on an ide channel I also had a chain of scsi > disks. I took out the scsi adapter alltogether thinking maybe there was > a conflict and hoping the pb might disappear but no, it's not that > either. I'm now on the verge of trying the the latest xfs release or the > latest kernel 2.4.18. > I took a look on the www.sgi.com site about xfs and there's nothing > about this, strange. Also SGI is a serious company which makes it hard > to believe the code would be so broken. It must be a combination of factors. > Greetings, > Guy. > > -- Bryan Whitehead SysAdmin - JPL - Interferometry Systems and Technology Phone: 818 354 2903 [EMAIL PROTECTED]
Re: [Cooker] Re: [expert] XFS and Linux-Mandrake
As I stated, I've seen the NULL or ^@ characters on systems that were shutdown proper. On Tue, 2002-04-02 at 16:39, Roger wrote: > On Tue, 2002-04-02 at 16:40, Bryan Whitehead wrote: > > On Tue, 2002-04-02 at 04:04, Guy Zelck wrote: > > > > > > You obviously haven't followed my lead : read the 'w > > > ich is better choice ext3 or ...' thread. A lot of detail is in there. > > > I can reproduce it any time by just doing un unclean shutdown or when I have to >when the system hangs. But maybe it has something to do with my SCSI card which is >not 100%. I'm getting closer. > > > > > > Guy. > > > > I should have mentioned I read the thread. We have had problems on > > machines that shutdown cleanly Files end up with null's in them > > instead of data. I can reproduce the error on basically any of our > > machines (around 20 Dell machines with onboard scsi). The problem never > > comes up with other FS's. (well, at least ext2) On very heavily loaded > > machines we've had entire directories disappear. Running any of the xfs > > tools does not restore missing directories or fix files that are full of > > null characters. > > > > If anyone would like instructions on how we trigger the problem about > > 30% of the time (for null characters), I can give detailed instructions > > and some scripts. It would be nice to have this resolved. We've had to > > switch back to ext2 since the problems were discovered. > > > > -- > > Bryan Whitehead > > SysAdmin - JPL - Interferometry Systems and Technology > > Phone: 818 354 2903 > > [EMAIL PROTECTED] > > > > I thought this problem was resolved in later releases of the XFS driver > (>2.4.13 kernel versions). I saw a noticable diff in stability with the > later kernel releases. > > Although, I haven't been performing unclean shutdowns here to test the > idea, but in earlier kernel versions, this was a real annoyance since I > would find allot of my user config files completely wiped :-( > > I've yet to see this problem persist in later kernel versions, although, > every now and then, i might come across an improperly formated config > file starting with , but moving further down in the file reveals the > config settings. I don't know if this is really XFS related or vim > related for sure. > > if you atleast do a alt+SysReq+S before a unclean shutdown, you should > be safe! this is the route of the problem/bug here. basically, the XFS > mailling list has more indepth detail about the sync bit > (err...whatever). > -- > Roger > - > Verify my pgp/gnupg signature on my HomePage: > http://www.alltel.net/~rogerx/about/index.html -- Bryan Whitehead SysAdmin - JPL - Interferometry Systems and Technology Phone: 818 354 2903 [EMAIL PROTECTED]
[Cooker] Re: [expert] XFS and Linux-Mandrake
On Tue, 2002-04-02 at 04:04, Guy Zelck wrote: > > You obviously haven't followed my lead : read the 'w > ich is better choice ext3 or ...' thread. A lot of detail is in there. > I can reproduce it any time by just doing un unclean shutdown or when I have to when >the system hangs. But maybe it has something to do with my SCSI card which is not >100%. I'm getting closer. > > Guy. I should have mentioned I read the thread. We have had problems on machines that shutdown cleanly Files end up with null's in them instead of data. I can reproduce the error on basically any of our machines (around 20 Dell machines with onboard scsi). The problem never comes up with other FS's. (well, at least ext2) On very heavily loaded machines we've had entire directories disappear. Running any of the xfs tools does not restore missing directories or fix files that are full of null characters. If anyone would like instructions on how we trigger the problem about 30% of the time (for null characters), I can give detailed instructions and some scripts. It would be nice to have this resolved. We've had to switch back to ext2 since the problems were discovered. -- Bryan Whitehead SysAdmin - JPL - Interferometry Systems and Technology Phone: 818 354 2903 [EMAIL PROTECTED]
[Cooker] Evolution 1.0.3 build problem
I can't seem to get evolution to build on a mandrake 8.1 system. rpm --rebuild evolution-1.0.3-1mdk.src.rpm gets to the "install" stage and craps out with this: make[1]: Leaving directory `/usr/src/RPM/BUILD/evolution-1.0.3' ++ cat /tmp/.X0-lock cat: /tmp/.X0-lock: No such file or directory + kill kill: usage: kill [-s sigspec | -n signum | -sigspec] [pid | job]... or kill -l [sigspec] error: Bad exit status from /var/tmp/rpm-tmp.78272 (%build) RPM build errors: Bad exit status from /var/tmp/rpm-tmp.78272 (%build) Anyone know what's going on? -- Bryan Whitehead SysAdmin - JPL - Interferometry Systems and Technology Phone: 818 354 2903 [EMAIL PROTECTED]
Re: [Cooker] ssh: end of discussion: MS make the right decision ?
> More to the point, the majority of us who do not want to find they have a > new system administrator should not get their machines broken into because > someone thought a service might be useful? > > Install it, yes. Enable it? Try it and your distro goes straight back to > the shop. Do NOT enable ANYTHING by default. I can't understand who your getting so bent out of shape over sshd? Linuxconf and webmin were not made by security freaks. sshd was. Telnet is obviously not a "secure" service. But they are all on. (in workstation) You said yourself (i think it was you) you use paranoid. We are talking about "Workstation". We want sshd in "Workstation". Sheesh. -- --- Bryan Whitehead Email: [EMAIL PROTECTED] WorkE: [EMAIL PROTECTED]
Re: [Cooker] ssh
> the problem is that you can't allow empty passwords in that case, big annoyance > for mostly home users, uh? > > maybe a solution would be a linux/webmin/??? function that would open sshd? As long as root has a password, the box *should* be ok. I personally remove linuxconf and webmin, they suck. They have there places in the home market and that's good / fine. But for the rest of the world < 200k on a 1.2GB install for sshd is not a lot to ask. If they select "Open to Crackers" then they know what they are getting into. If they use WinXX they also know what they are getting into. :) -- --- Bryan Whitehead Email: [EMAIL PROTECTED] WorkE: [EMAIL PROTECTED]
