Help - Strangeness in a Redhat
Hey ppl, I have a couple of strange things happening on my system, if anyone has any ideas on what is cause them please let me know. First, I am running RH 9 on my workstation and have a RH 7.1 box running as a webserver. I mount /home via nfs from the RH 7 box to my workstation. I have noticed that write speeds to /home are very slow, with no obvious reason. If I copy a file from /tmp (local) to /home/raymond (remote) it takes forever, and there is next to no network usage or cpu load. Transfers in the opersite direction saturate the 10MBit connection. I am using the following options for mount: rsize=8192,wsize=8192,intr Any Ideas why writes would be so slow? I have ftp running on the RH 7 box and ftping the data to /home/raymond via ftp runs at fullspeed. Second issue is a also strange, once I reboot I am able to run up2date and any rpm command, so I can install packages but after a day or two for no reason these commands just stop. They hang on execution as if they are waiting for something. You can't stop them with CTRL-C either, you have to kill -9 the process. This means I have to reboot my machine if want to install a new package, which isn't as much of a problem now I have things installed but was initially a pain in the bu when I first built the machine. Any Idea? Thanks, Raymond.
Second issue is a also strange, once I reboot I am able to run up2date and any rpm command, so I can install packages but after a day or two for no reason these commands just stop. They hang on execution as if they are waiting for something. You can't stop them with CTRL-C either, you have to kill -9 the process. This means I have to reboot my machine if want to install a new package, which isn't as much of a problem now I have things installed but was initially a pain in the bu when I first built the machine.
Your problem is described at https://bugzilla.redhat.com/bugzilla/show_bug.cgi?id=73097 and apparently fixed by a later version of rpm. Red Hat haven't released an errata for it, but you can get it out of Rawhide. The 'solution' is, when you've killed rpm, remove the stale locks at /var/lib/rpm/__db.* You may also have to rpm --rebuilddb afterwards, but I've never had to. Craig
Raymond Burgess wrote:
Hey ppl,
I have a couple of strange things happening on my system, if anyone has any ideas on what is cause them please let me know.
First, I am running RH 9 on my workstation and have a RH 7.1 box running as a webserver. I mount /home via nfs from the RH 7 box to my workstation. I have noticed that write speeds to /home are very slow, with no obvious reason. If I copy a file from /tmp (local) to /home/raymond (remote) it takes forever, and there is next to no network usage or cpu load. Transfers in the opersite direction saturate the 10MBit connection.
I am using the following options for mount: rsize=8192,wsize=8192,intr
I had this probably on a machine a couple of computers ago and I'm not sure exactly of the details but I think I had to change the wsize to something smaller and that sped things up. I think it had something to do with being compatible with the buffer size on the enet card. I think I may have read about that in the NFS HOWTO. g -- Glenn Ramsey <glenn(a)componic.co.nz> 07 8627077 http://www.componic.co.nz
Glenn Ramsey wrote:
Raymond Burgess wrote:
Hey ppl,
I have a couple of strange things happening on my system, if anyone has any ideas on what is cause them please let me know.
First, I am running RH 9 on my workstation and have a RH 7.1 box running as a webserver. I mount /home via nfs from the RH 7 box to my workstation. I have noticed that write speeds to /home are very slow, with no obvious reason. If I copy a file from /tmp (local) to /home/raymond (remote) it takes forever, and there is next to no network usage or cpu load. Transfers in the opersite direction saturate the 10MBit connection. I am using the following options for mount: rsize=8192,wsize=8192,intr
I had this probably on a machine a couple of computers ago and I'm not sure exactly of the details but I think I had to change the wsize to something smaller and that sped things up. I think it had something to do with being compatible with the buffer size on the enet card.
I think I may have read about that in the NFS HOWTO.
g
-- Glenn Ramsey <glenn(a)componic.co.nz> 07 8627077 http://www.componic.co.nz
Glenn Ramsey wrote:
Raymond Burgess wrote:
Hey ppl,
I have a couple of strange things happening on my system, if anyone has any ideas on what is cause them please let me know.
First, I am running RH 9 on my workstation and have a RH 7.1 box running as a webserver. I mount /home via nfs from the RH 7 box to my workstation. I have noticed that write speeds to /home are very slow, with no obvious reason. If I copy a file from /tmp (local) to /home/raymond (remote) it takes forever, and there is next to no network usage or cpu load. Transfers in the opersite direction saturate the 10MBit connection. I am using the following options for mount: rsize=8192,wsize=8192,intr
I had this probably on a machine a couple of computers ago and I'm not
That "probably" should say "problem", sorry about that. And sorry about the quote only post just now. Had too many beers to be doing this (and the the sad thing is that's only 2).
sure exactly of the details but I think I had to change the wsize to something smaller and that sped things up. I think it had something to do with being compatible with the buffer size on the enet card.
I think I may have read about that in the NFS HOWTO.
g
-- Glenn Ramsey <glenn(a)componic.co.nz> 07 8627077 http://www.componic.co.nz
First, I am running RH 9 on my workstation and have a RH 7.1 box running as a webserver. I mount /home via nfs from the RH 7 box to my workstation. I have noticed that write speeds to /home are very slow, with no obvious reason. If I copy a file from /tmp (local) to /home/raymond (remote) it takes forever, and there is next to no network usage or cpu load. Transfers in the opersite direction saturate the 10MBit connection.
I am using the following options for mount: rsize=8192,wsize=8192,intr
I had this probably on a machine a couple of computers ago and I'm not sure exactly of the details but I think I had to change the wsize to something smaller and that sped things up. I think it had something to do with being compatible with the buffer size on the enet card.
I think I may have read about that in the NFS HOWTO.
Cheers Glenn, I have removed the {r,w}size from the options and using the default read and write sizes, 4096 I think. Makes a big difference to writes. The strange thing was there appeared to be no traffic on the device when it was running slow. I have been running happily with {r,w}size values of 8192 for many years. I have released now that this is the first time in many years I have been running with a Half-Duplex hub in the network. Raymond
Try an rpm --rebuilddb to fix the RPM issues.
First, I am running RH 9 on my workstation and have a RH 7.1 box running as a webserver. I mount /home via nfs from the RH 7 box to my workstation. I have noticed that write speeds to /home are very slow, with no obvious reason. If I copy a file from /tmp (local) to /home/raymond (remote) it takes forever, and there is next to no network usage or cpu load. Transfers in the opersite direction saturate the 10MBit connection.
Are you getting any TX or RX errors on the wire? Collisions etc? Retransmissions? Regards -- Oliver Jones » Director » oliver.jones(a)deeperdesign.com » +64 (21) 41 2238 Deeper Design Limited » +64 (7) 377 3328 » www.deeperdesign.com
First, I am running RH 9 on my workstation and have a RH 7.1 box running as a webserver. I mount /home via nfs from the RH 7 box to my workstation. I have noticed that write speeds to /home are very slow, with no obvious reason. If I copy a file from /tmp (local) to /home/raymond (remote) it takes forever, and there is next to no network usage or cpu load. Transfers in the opersite direction saturate the 10MBit connection.
Are you getting any TX or RX errors on the wire? Collisions etc? Retransmissions?
No notable transmission errors, there are a few errors and collisions, but nothing out of the ordinary for a Half Duplex 10 Mbit link. The Network is fine as I am able to FTP the file at full speed.
participants (4)
-
Craig Box -
Glenn Ramsey -
Oliver Jones -
Raymond Burgess