Re: [Cooker] NFS install troubles
On Mon, 23 Apr 2001 21:41:22 -0400 Pierre Fortin <[EMAIL PROTECTED]> [Re: [Cooker] NFS install troubles] wrote : > Eaon wrote: > > > >What a lazy bum I can be on > > weekends. But something else jumped out at me this weekend. I was > reading > > the latest ReiserFS sucks/rocks flame war on linuxtoday.com and one > person > > mentioned that there were problems with NFS working with Reiser. > Could that > > have something to do with it? The partition on the NFS server > (Mandrake > > 7.2) that I am trying to mount is Reiser. Any issues there that you > (or > > anyone, feel free to jump in) know of? > > This is a possibility since your server is running Reiser; the NFS > crashing > problem is a separate issue that occurs with ext2 though it's possible > that they > are fstype independent. > The problem with ReiserFS and NFS is from a quota POV. They are conceptualy incompatible. And without any tricks it's impossible to get ReiserFS + NFS + quota. Anyway ReiserFS + NFS is more than a Stephen King nightmare ! my 20cents. > > Eaon > > Pierre > > -- + Fabien Lebaillif Delamare R&D Engineer + Office : +65 844 1301 HP : +65 9848 1431 + 18 Tannery Lane #03-05 Lian Tong Building Singapore 347780 +
Re: [Cooker] NFS install troubles
Eaon wrote: > >What a lazy bum I can be on > weekends. But something else jumped out at me this weekend. I was reading > the latest ReiserFS sucks/rocks flame war on linuxtoday.com and one person > mentioned that there were problems with NFS working with Reiser. Could that > have something to do with it? The partition on the NFS server (Mandrake > 7.2) that I am trying to mount is Reiser. Any issues there that you (or > anyone, feel free to jump in) know of? This is a possibility since your server is running Reiser; the NFS crashing problem is a separate issue that occurs with ext2 though it's possible that they are fstype independent. > Eaon Pierre
RE: [Cooker] NFS install troubles
> Dunno. I'm too busy to really dig into the sources for various > problems I've > analyzed and reported; probably by the time 8.1 comes out, I'll > be able to spend > more time with the sources. So much for retirement... :^) > If there's something as cool as Linux to play with when I retire, I won't be complaining. ;-) > Anyway, did you have a look at the /var/log/kernel/warnings[*gz] > files to see if > you had any nfsd crashes? If you did, send me the crash data > therein and I'll > keep them for analysis as more a reported... > > To figure out why the connections fail, I find ethereal to be a > great asset... > If you get a trace and don't understand it, feel free to save the > data and send > it to me. > I haven't had a chance to do any of this. What a lazy bum I can be on weekends. But something else jumped out at me this weekend. I was reading the latest ReiserFS sucks/rocks flame war on linuxtoday.com and one person mentioned that there were problems with NFS working with Reiser. Could that have something to do with it? The partition on the NFS server (Mandrake 7.2) that I am trying to mount is Reiser. Any issues there that you (or anyone, feel free to jump in) know of? Eaon
Re: [Cooker] NFS install troubles
Mandrake 7.2 server On 21 Apr 2001 21:36:45 +0200, Juan Quintela wrote: > > "eaon" == Eaon <[EMAIL PROTECTED]> writes: > > Hi > > eaon> When I supply the IP (all IPs are static) and path to where the files > eaon> are located, then hit enter, I can see on the hub that the lights for > eaon> the client and server machines are blinking together (and that's the > eaon> only network traffic there is), so there is some form of communication, > eaon> but they blink about 5 or 6 times then it tells me it couldn't mount the > eaon> directory. There is no error on the logs console on the client (no "RPC > eaon> Timeout" or anything, just the "preparing nfsmount" line, that's the > eaon> last thing). > > What system is running the NFS server? It happened to me that with a > debian woody (unstable) NFS server, I had a lot of NFS traffic on the > wire, but the installer don't show anything. As I was in a hurry and > the machine was a unstable server I didn't investigate more about > that. Could you confirm what system is running in your server please? > > Later, Juan. > > -- > In theory, practice and theory are the same, but in practice they > are different -- Larry McVoy >
Re: [Cooker] NFS install troubles
> "eaon" == Eaon <[EMAIL PROTECTED]> writes: Hi eaon> When I supply the IP (all IPs are static) and path to where the files eaon> are located, then hit enter, I can see on the hub that the lights for eaon> the client and server machines are blinking together (and that's the eaon> only network traffic there is), so there is some form of communication, eaon> but they blink about 5 or 6 times then it tells me it couldn't mount the eaon> directory. There is no error on the logs console on the client (no "RPC eaon> Timeout" or anything, just the "preparing nfsmount" line, that's the eaon> last thing). What system is running the NFS server? It happened to me that with a debian woody (unstable) NFS server, I had a lot of NFS traffic on the wire, but the installer don't show anything. As I was in a hurry and the machine was a unstable server I didn't investigate more about that. Could you confirm what system is running in your server please? Later, Juan. -- In theory, practice and theory are the same, but in practice they are different -- Larry McVoy
Re: [Cooker] NFS install troubles
Eaon wrote: > > > Could be a server problem... On the server, what do you get from: > > > >netstat -l | grep 2049 > > > > If you see someting like this: > > > >udp16992 0 *:2049 *:* > > ^ (anything but zero) > > retry the command; if still the same, get the nfsd crash data from > > /var/log/kernel/warnings and send it in. > > > > I've already reported that NFS sometimes crashes and only > > solution I can find > > requires server reboot. I still have LM6.1 and LM7.2 machines in > > that state if > > Mandrake want to get more info (what logrotate hasn't removed). > > > > Pierre > > > > Pierre, > > I tried this last night (after the Oilers/Stars game, of course), and wasn't > able to get the 7.2 NFS server to crash again (why is it they never crash > when you want them to?), but still wasn't able to connect from the network > install disk from cooker. > > Eaon Dunno. I'm too busy to really dig into the sources for various problems I've analyzed and reported; probably by the time 8.1 comes out, I'll be able to spend more time with the sources. So much for retirement... :^) Anyway, did you have a look at the /var/log/kernel/warnings[*gz] files to see if you had any nfsd crashes? If you did, send me the crash data therein and I'll keep them for analysis as more a reported... To figure out why the connections fail, I find ethereal to be a great asset... If you get a trace and don't understand it, feel free to save the data and send it to me. Pierre -- Support Linux development: http://www.linux-mandrake.com/donations/ Last reboot reason: 01/03/27: winter storm 6hr power outage
RE: [Cooker] NFS install troubles
> Could be a server problem... On the server, what do you get from: > >netstat -l | grep 2049 > > If you see someting like this: > >udp16992 0 *:2049 *:* > ^ (anything but zero) > retry the command; if still the same, get the nfsd crash data from > /var/log/kernel/warnings and send it in. > > I've already reported that NFS sometimes crashes and only > solution I can find > requires server reboot. I still have LM6.1 and LM7.2 machines in > that state if > Mandrake want to get more info (what logrotate hasn't removed). > > Pierre > Pierre, I tried this last night (after the Oilers/Stars game, of course), and wasn't able to get the 7.2 NFS server to crash again (why is it they never crash when you want them to?), but still wasn't able to connect from the network install disk from cooker. Eaon
RE: [Cooker] NFS install troubles
> > and HTTP require 52 MB - who picked that number?? ;-) ), and a first > > We. Ramdisk takes RAM. > Cool. You got 4 meg of RAM, a soldering iron and a hammer? I think that should do it to get this system up to spec. ;-) > Are you sure you have nothing on log console? > Absolutely positive. Even got it to a point where NFS wasn't working for anything (more on this later), and on the cooker machine I get "mount: RPC: Timeout", but definitely nothing in the install. Just the message on console 1 "I couldn't mount the directory" (or whatever it says). > Also, pleasy try to nfs mount from a valid host. Here's where I thought you had me! "Ah," I thought, "I'm a moron". I checked the exports file and it turns out the IP I picked for this machine didn't have rights. So I redid the exports so there were no restrictions on the export, and now NFS is just hooped on the server machine. This happens often. It's a 7.2 machine, and after a reboot or after I change something in exports (usually through linuxconf) and reload NFS, it often just stops working. This morning after I redid the export through linuxconf it wouldn't work. Couldn't even mount from the local machine to itself. So I rebooted (I think I need a 12 step program to help end that M$-induced habit), and was able to mount locally on the 7.2 machine to itself. Tried it from the cooker machine, and it timed out. Same from the install. Went back to the 7.2 machine and tried again, and it timed out too. Something crashed it? Or a better question: What is it with NFS on 7.2? ;-) Then I had to leave for work (where I write this - shhh!), so I couldn't play anymore. > This nfs mount stuff > should do it. Please also try to have a 2.4 kernel on the nfs server, > there are some weird incompatibilities between different kernels for > server and client sometimes. > I thought of that. The 7.2 machine is the stock 2.2.17-21mdk kernel. But the cooker machine is 2.4.3-5mdk, and I tried exporting something from that machine (not doing anything stupid like restricting access) and directing the installer to that, and that wouldn't mount either. But I didn't try that one too hard, so I could have messed that up too. > They don't have a graphical install -- do they? ;-) > No, they have an evil install that makes you partition and format the hard drive before you do anything else at all, including setting up your network card, so if your NIC isn't BSD compatible, you don't get to find out until you've completely trashed the system. I like the Mandrake way better for that. :-) > Please try with an alternatives/network.img to have a 2.2.19 kernel, so we > can understand if it's incompatibility between kernels. > > > Oh, ya, and I tried all three network boot disks (the main network.img > > and the two "alternatives"), and it didn't help any. > > Argh. Well. > Ya, that's what I said too. > -- > Guillaume Cottenceau - http://us.mandrakesoft.com/~gc/ > Eaon
Re: [Cooker] NFS install troubles
Eaon wrote: > > Trying to install via NFS and it just won't do it. [...] > When I supply the IP (all IPs are static) and path to where the files > are located, then hit enter, I can see on the hub that the lights for > the client and server machines are blinking together (and that's the > only network traffic there is), so there is some form of communication, > but they blink about 5 or 6 times then it tells me it couldn't mount the > directory. There is no error on the logs console on the client (no "RPC > Timeout" or anything, just the "preparing nfsmount" line, that's the > last thing). [...] > Ideas? Could be a server problem... On the server, what do you get from: netstat -l | grep 2049 If you see someting like this: udp16992 0 *:2049 *:* ^ (anything but zero) retry the command; if still the same, get the nfsd crash data from /var/log/kernel/warnings and send it in. I've already reported that NFS sometimes crashes and only solution I can find requires server reboot. I still have LM6.1 and LM7.2 machines in that state if Mandrake want to get more info (what logrotate hasn't removed). > Eaon Pierre -- Linux (Up 28 days) -- Reboots are for system upgrades... not Windows X^P Last reboot reason: 01/03/27: winter storm 6hr power outage
Re: [Cooker] NFS install troubles
Eaon <[EMAIL PROTECTED]> writes: > Trying to install via NFS and it just won't do it. IBM Aptiva C30 (if > that means anything to anyone - a friend gave it to me for free because > he got a job in the Caribbean and couldn't take it with him - I cried > for him, really, but I promised to give the system a good home). > Pentium 233, 3c509 ISA NIC, 48 MB RAM (so I have to use NFS, since FTP > and HTTP require 52 MB - who picked that number?? ;-) ), and a first We. Ramdisk takes RAM. > generation DVD drive that steadfastly refuses to read burned CDs, > leaving NFS as the only way to get the code in there. > > When I supply the IP (all IPs are static) and path to where the files > are located, then hit enter, I can see on the hub that the lights for > the client and server machines are blinking together (and that's the > only network traffic there is), so there is some form of communication, > but they blink about 5 or 6 times then it tells me it couldn't mount the > directory. There is no error on the logs console on the client (no "RPC > Timeout" or anything, just the "preparing nfsmount" line, that's the > last thing). I pretty sure that the hardware is good - the system had Strange. I'm pretty sure you should have at least one of the following error message: NFS server says: No such file or directory NFS server says: Permission denied nfsmount prepare failed, reason given by server: %d mount failed: %s Are you sure you have nothing on log console? Also, pleasy try to nfs mount from a valid host. This nfs mount stuff should do it. Please also try to have a 2.4 kernel on the nfs server, there are some weird incompatibilities between different kernels for server and client sometimes. > Win95 on it when I got it (like I said, I told him I'd treat it well, so > to keep that promise, Win95 had to go) and I was able to browse the net, > and in a fit of frustration when Mandrake wouldn't work I installed > OpenBSD via FTP (they don't have a 52 meg limit ;-) ), connecting to They don't have a graphical install -- do they? ;-) > the same server machine as I tried for NFS, and that all worked. From > the server side, I was able to mount the cooker location from a third > machine via NFS, so I am pretty sure that is working too. The server is > LM7.2. The "third location" machine is another cooker install. Please try with an alternatives/network.img to have a 2.2.19 kernel, so we can understand if it's incompatibility between kernels. > Oh, ya, and I tried all three network boot disks (the main network.img > and the two "alternatives"), and it didn't help any. Argh. Well. -- Guillaume Cottenceau - http://us.mandrakesoft.com/~gc/