Re: [CentOS] USB drive fails at sector 0xFFFFFFF
On Sun, Sep 7, 2008 at 4:44 PM, Kenneth Porter [EMAIL PROTECTED] wrote: I'm backing up to a NTFS partition on an external USB drive with dump. I'm seeing failures in /var/log/messages reading sector 0xFFF that cause the verify pass to fail. Are there any known problems in the USB driver? Kernel via uname -a: Linux segw2.mpa.lan 2.6.18-92.1.6.el5 #1 SMP Wed Jun 25 13:49:24 EDT 2008 i686 i686 i386 GNU/Linux Message reported. (Note the number 268435455, which is 0xFFF, and occurs every time I see this happen. You are backing up to NTFS on Linux? The NTFS file system specification is a trade secret last I checked.. Anytime I see an error where the number is binary all ones I suspect a magic number... and worry. How was the disk partitioned? How was this partition formatted? Was the filesystem converted from old to new ntfs format? http://technet.microsoft.com/en-us/library/cc781134.aspx -- NiftyCluster T o m M i t c h e l l ___ CentOS mailing list CentOS@centos.org http://lists.centos.org/mailman/listinfo/centos
Re: [CentOS] Problem with running Centos 5.2 on Dell Optiplex 330
On Mon, Sep 8, 2008 at 2:41 AM, Niki Kovacs [EMAIL PROTECTED] wrote: NiftyClusters Mitch a écrit : The quick and handy way to do this is to pick up an inexpensive USB ethernet link or other ether net card. After a yum update to a new kernel you may well have a driver that works. 'yum update' without a working network card won't lead him far, IMHO :o) You missed the inexpensive USB ethernet link part. -- NiftyCluster T o m M i t c h e l l ___ CentOS mailing list CentOS@centos.org http://lists.centos.org/mailman/listinfo/centos
Re: [CentOS] check mal server speed n performance
On Sun, Sep 7, 2008 at 9:33 AM, fabian dacunha [EMAIL PROTECTED] wrote: Dear All, I have Centos 5.1 and sendmail mail server running for over a year and been workin fine i also use mailscanner + clamav now some remote users have been complaing that the mail sometimes time out i have even tried to stop the mail scanner and jus test my sendmail but the problem exists now i would apprecite if someone cd help me or let me know of a link where i could test the mail server performance and know if its working fine so to rule out the possibility of any problem with my mail server software the mail logs r normall and dont show any different thing thanks and regards apprecite ur help The most likely issues are DNS, virus scan and yet one too many messages. Look hard at DNS resolution... spam filters now do host name checks forward, back, inside, out. A volume email server needs access to a local quick name server. Adding a local name server or moving the local name server to a dedicated machine could be a win. There is no easy way to benchmark a mail server that is in operation. One thing to tinker with is a fetchmail benchmark of a known set of messages from some place like gmail. You can watch the mail get processed (mailq), watch the name server and file IO etc. By refilling your INBOX from All Mail or a saved folder you should be able to play it again. If you need a pile of spam use a throw away gmail account and almost register for one of those free laptop offers that Google pops up. -- NiftyCluster T o m M i t c h e l l ___ CentOS mailing list CentOS@centos.org http://lists.centos.org/mailman/listinfo/centos
Re: [CentOS] Problem with running Centos 5.2 on Dell Optiplex 330
On Sun, Sep 7, 2008 at 7:29 AM, nate [EMAIL PROTECTED] wrote: Pete Kay wrote: Hi, I am havind deep trouble with a bunch of our newly arrived Optiplex 330 as it can't run Centos 5.2 property. .. Does anyone know how I can fix this problem? It's quite possible the network driver isn't compatible with the system. Looking at the support site for Dell that system isn't supported with Linux, and it looks like it has a pretty new broadcom chip on it. I'd bet that updating the driver will allow the NIC to work, the latest drivers for that system are available. The quick and handy way to do this is to pick up an inexpensive USB ethernet link or other ether net card. After a yum update to a new kernel you may well have a driver that works. If a yum update does not update the driver, you have enough connectivity to download then install vendor bits and tools. This is the trick I have used for a number of early access Intel 'engineering' boxes. It lets me kickstart to a lab configuration and connect long enough to add the updated driver. I can save all the fun new bits and script up the driver update into a single dir to copy from one to the next. Then burn that dir of bits to a CDROM... If you are shopping, the older the USB ethernet device the more likely it is to work. -- NiftyCluster T o m M i t c h e l l ___ CentOS mailing list CentOS@centos.org http://lists.centos.org/mailman/listinfo/centos
Re: [CentOS] DJB's daemontools package
Sorry this is slightly off topic. On Sat, Sep 6, 2008 at 6:55 AM, Jeff Kinz [EMAIL PROTECTED] wrote: On Sat, Sep 06, 2008 at 05:51:15AM -0700, Al Sparks wrote: Oh and the qmail server? My employer went Exchange. And slowly but surely, the IT there is becoming more Microsoft with Linux becoming more of an outlier. It's probably time for me to find another job. It's hard, because I've been with them a long time. === Al Or its time for you to make a direct presentation to the upper level management showing that the number of direct staffers needed to administer an MS environment is between 15 and 300% higher than a corresponding UNIX/Linux environment. Why Upper management and not IT management? Because IT management BENEFITS from having more staff to manage. When your headcount increases and your budget is larger, your management importance and status goes up and you become recognized as a peer of those who may become the next CEO. UNIX/Linux is bad for the IT empire builders because it is good for the rest of the company. It reduces costs and headcount while increasing increases reliability and security at the same time. Jeff Kinz Jeff is correct If you look at the IT manpower and how the org charts play. The windows support staff will map out to more managers and more managers of managers. The side effect is that the audience of any presentation will be level mismatched. Windows oriented managers will be pitching budget and needs two or three levels above a Linux support staff doing the same work simply because the staff size mismatch. At a previous Unix based hardware computer company I did some curiosity driven research and noted that the company had more staff (mostly contractors) supporting windows for the internal needs of the company than the company had support staff (both hardware and software) for the paying customers. The organization structure was disjoint and difficult to map At times a windows support person would show up to fix something I would look at the name tag and try to find what group the person was in and most often I would not have considered it a windows support group. Much of the Unix support was done in the spare time by a handful of folk where the Windows support was managed by a 'staff'. Very different staffing and management model Also there is the comfort factor -- schools teach Windows stuff in computer science departments. Managers and sales staff that can type know outlook, excel and word. To complicate this, legal requirements mandate lots of stuff. In the 'free' software universe we live in it is hard to comprehend how much money is spent to guarantee that all software is legal. Compliance in this regard costs a lot. Of interest the Unix community and the Windows world have very different security models. For the most part Unix is transparent while windows is opaque. There is a reason that VMS was so popular -- NiftyCluster T o m M i t c h e l l ___ CentOS mailing list CentOS@centos.org http://lists.centos.org/mailman/listinfo/centos
Re: [CentOS] Yum corrupting RPMs
On Sun, Aug 24, 2008 at 6:24 PM, Tom Lanyon [EMAIL PROTECTED] wrote: Hi list, Trying to upgrade someone's workstation here to 5.2 (was installed from a 5.0 DVD I think). The RPMs on our internal mirror are in-tact and pass a 'rpm --checksig' test, yet when I run a 'yum upgrade' a large portion of them are corrupted and fail the GPG check. Since yum has a local cache you may need to invoke one of the clean flags for yum. * clean [ packages | headers | metadata | dbcache | all ] If you do a clean all we will not know which of the set is bogus... My bet is the dbcache with metadata to show. Given what we now know, it might be good to copy the cache of packages and do a local compare of any that are freshly downloaded. Since yum depends on RPM you should be able to download individual packages and then inspect each with RPM tools as you are doing... -- NiftyCluster T o m M i t c h e l l ___ CentOS mailing list CentOS@centos.org http://lists.centos.org/mailman/listinfo/centos
Re: [CentOS] Problem copying files to flash drive
On Mon, Aug 18, 2008 at 12:41 PM, MHR [EMAIL PROTECTED] wrote: I had an interesting experience this weekend backing up some flash drives to another flash drive on my CentOS 5.2 home desktop. My son had two 256MB flash drives and one 1GB flash drive that he wanted backed up onto his newer 2GB flash drive. I used rsync to copy the two smaller ones to the big one without any trouble, but when I tried to backup the 1GB files to the 2GB drive, I started getting massive no space on disk errors. After doing this several times, I copied to files to a directory on my hard drive and then tried to copy the files again from there to the 2GB flash drive (removing all files on the flash drive in between each iteration). Same problem. Then I did a cp -R of the directory itself from my hard drive to the flash drive, and voila! All files copied, no problems, except that now the files are in a new directory of the same name on the flash drive. The total number of files in the base directory is 171, and the total size of the files copied was a little over 500MB, so I believe that everything should have fit just fine. The commands I used that failed were 'rsync -av /media/drive_name/ /media/2GB_drive_name/' (which worked for the two smaller drives but not the bigger one), and then (after using the same command to copy the files to my hard drive), 'rsync -av ./ /media/2GB_drive_name/' The command that worked was 'cp -R . /media/2GB_drive_name/' The only thing that comes to mind is that there were a lot of files with ._xxx names on the 1GB flash drive, so I'm wondering if they just overloaded the capacity of the flash drive's FAT32 root directory name space, but I thought that limit was 512 entries, not less than 171. Any ideas? Thanks. mhr Sparse files?? See the -S or --sparse flag -- NiftyCluster T o m M i t c h e l l ___ CentOS mailing list CentOS@centos.org http://lists.centos.org/mailman/listinfo/centos
Re: [CentOS] File system goes read-only once in a while
On Fri, Aug 1, 2008 at 1:43 PM, William L. Maltby [EMAIL PROTECTED] wrote: On Fri, 2008-08-01 at 16:13 -0400, Toby Bluhm wrote: Mufit Eribol wrote: snip . . that you would correctly try to fsck the *device*. First backup data... It is possible to run fsck with a media test flag. Bad blocks are assigned to dummy files. Inadvertently reading one of these files can take a drive off line. One reason a device will go off line is the presence of a media error, or the presence of a situation assumed by smartd to be a pending data risk. Understanding the root cause error should be done. Smartd tends to be cautious but does identify pending problems. One puzzle can be the loss of log file data. It is sometimes possible to see events on a live system that later vanish after a reboot because buffers are live in memory but not on the disk. Sending logs to another 'log system' can be helpful and is a good idea on production systems for exactly this reason. -- NiftyCluster T o m M i t c h e l l ___ CentOS mailing list CentOS@centos.org http://lists.centos.org/mailman/listinfo/centos
Re: [CentOS] Port Closing Question..
On Thu, Jul 31, 2008 at 2:06 PM, Ryan Nichols [EMAIL PROTECTED] wrote: Is there a software avail or a process that will monitor two ports and if there is no traffic close them so the program that is using them can reuse them? I talked to the vendor and they told me I needed to do this on the NAT/Firewall , but I dont see anything like that on my router. So any suggestions ideas? thanks, Ryan Nichols Do some historic searching on WEB server mailing lists. Web servers have a classic problem where the close of a socket is a multiple step process involving both ends of the socket. Killing the process (kill -9) or even disconnecting from a wireless hot spot can leave the server system with an open socket that could last for days In the old days web servers would have to be rebooted to clear this. There is now a way to drop connections after a timer expires. It is a kernel param and some think it violates the TCP/IP specification ;-) so the default is off or very long. Slightly different are NFS, ssh and rsh links that hang out until the far system comes back. But both are a failure for the cooperative shut down of the link. I am not sure a common simple firewall/ NAT box can help this Cisco and others have some good documents on line... see also CLOSE_WAIT and FIN in the TCP/IP specification... and look for related Linux network flags -- NiftyCluster T o m M i t c h e l l ___ CentOS mailing list CentOS@centos.org http://lists.centos.org/mailman/listinfo/centos
Re: [CentOS] Ethernet poor performance
Good list: Also add multiple runs of traceroute and also try ping, ping -f , ping -A and ping -R. See also ping6 If routes are dynamic we have one answer to the problem, I would expect traceroute to have 'one' answer on a simple net. If packets are falling on the floor then we need to know why. The different invocations of ping can tell you if packets drop at slow or fast transfer rates. ping -R is slightly different than traceroute but if the return routes flip one way then the other we should know why. Watch out for /etc/sysconfig/network-scripts/ifcfg* where hardware addresses, subnet masks, device driver links/ names etc. no longer match the hardware when things move. Consider snooping packets on the link to see if all is as you expect. I keep an old, slow network hub (not a switching hub) for the times when I want to see the bits on the wire and not the bits that the local driver is able to show me. On Wed, Jul 2, 2008 at 8:55 AM, nate [EMAIL PROTECTED] wrote: Robert Moskowitz wrote: Any Idea where I should look to fix this? - What version of CentOS? - What type of network card? - What driver is it using? - What type of device is on the other end of the network card?(Switch, hub, router etc) - Can you verify that the speed and duplex settings match on both ends of the connection? - I assume the ping you are running is only 1 or 2(if there's a hub/switch in between) hops away? - Any errors reported by ifconfig ? any collisions? - Try replacing the card itself? maybe it is bad. nate ___ CentOS mailing list CentOS@centos.org http://lists.centos.org/mailman/listinfo/centos -- NiftyCluster T o m M i t c h e l l ___ CentOS mailing list CentOS@centos.org http://lists.centos.org/mailman/listinfo/centos
Re: [CentOS] RHEL vs. CentOS
Exactly, RedHat service (not free) uses the redhat update network and tools like up2date to deliver updated RPM packages to their customer. They do not distribute yum by default and they do not have a yum repository configuration that works. Under the covers both RHEL and CentOS use RPMs so rpm -Uvh this.rpm that.rpm works for both. Because of this yum can be installed and a local yum repository can be used for local packages. RPM will do the right thing as long as the local packages are 'well' made. From the support point of view expect redhat updates to be delivered to RH customers a min. to a week sooner than the CentOS community sees them. Start using language like package update tool where you would commonly use yum. Scan the online documentation for redhat network, up2date, rhn etc In with the exception of rhn .vs. yum and copyright images there is no difference beyond possible random differences (minor) in the hosts that RPMs are built on. Also /etc/redhat-release will (OK should) differ. On Wed, Jul 2, 2008 at 9:09 AM, Johnny Hughes [EMAIL PROTECTED] wrote: drew einhorn wrote: Hi, we have a new customer to support. They have RHEL5 not CentOS5. Is there a summary of the differences between RHEL and CentOS? We have come across differences in how yum works. There is magic involving the rhnplugin that connects us to invisible repositories not mentioned in the yum config files, interacts with the web interface via rhn.redhat.com, ... We wonder if there are other differences that we have not yet stumbled upon, that might create issues in the future. Your client can call red hat support and there is a working yum-security plugin. They also have access to RHN to manage assets. That is about it for differences. You need to think more about things like compiling custom kernels or 3rd party repos, since you have upstream paid support now. ___ CentOS mailing list CentOS@centos.org http://lists.centos.org/mailman/listinfo/centos -- NiftyCluster T o m M i t c h e l l ___ CentOS mailing list CentOS@centos.org http://lists.centos.org/mailman/listinfo/centos