Wilton wrote: > > >>> I've been running a CentOS server connected to a Promise VTrak m500i >>> via open-iscsi quite successfully for over a year now. A few days ago >>> the performance plummeted on the drive and I'm seeing the following >>> messages in my log file: >>> Jan 14 09:06:13 flax kernel: iscsi-sfnet:host7: ping timeout of 5 secs >>> expired, last rx 161502341, last ping 161507 >>> 341, now 161512341 >> What kernel and open-iscsi version are you using? This looks like >> something from the open-iscsi.org test release (also in scsi-misc and in >> the future RHEL 5.1 kernel). > > I'm running CentOS 4.5.
> >> Did you just upgrade the kernel or tools? > > No. That's what's weird. It's been running fine for months and then Ah sorry. For some new open-iscsi code I copied the linux-iscsi code in Centos/RHEL 4.5. And I used the same error message as the Centos code, so that threw me. Ignore me above. > all of a sudden, last Thursday, these messages started showing up. I > thought it might be a problem with the switch so about an hour ago I > moved all the ethernet cabling to alternative ports on the switch, but > that didn't resolve the problem. > >> Ever X seconds the initiator will send a nop to check the target. You >> can control this with the noop settings in /etc/iscsi/iscsi.conf. If it >> does not get a response it will drop the session and retry commands. It >> looks like this happened here. >> >>> Jan 14 09:06:13 flax kernel: iscsi-sfnet:host7: Session dropped >>> Jan 14 09:06:13 flax kernel: iscsi-sfnet:host7: Failing command cdb >>> 0x28 task 264952 with return code = 0x20000 >>> <snip> >>> Jan 14 09:06:13 flax kernel: iscsi-sfnet:host7: Failing command cdb >>> 0x28 task 264959 with return code = 0x20000 >>> Jan 14 09:06:13 flax kernel: iscsi-sfnet:host7: Failing command cdb >>> 0x8a task 264960 with return code = 0x20000 >>> Jan 14 09:06:13 flax kernel: iscsi-sfnet:host7: Waiting 2 seconds >>> before next login attempt >>> Jan 14 09:06:15 flax kernel: iscsi-sfnet:host7: Session established >> Then we were able to log back in right away. >> >> So either there was a problem at the target/network and we did not get a >> iscsi ping response, or if you just upgraded to some new code and are >> not seeing it maybe there is a bug in there. > > I'll look into this noop stuff. The thing is, I can do a flood ping > from the host to the Promise array and it doesn't lose any packets at > all. Why would the iscsi miss a packet. > I do not know. We may have hit a problem where the cmdsn numbering is wrong. open-iscsi had messed that up and promise targets did not like it, but when I looked at linux-iscsi/RHEL/Centos it looked ok. Maybe I am thinking about something else. A wireshark trace or the target should have logged a error. > And, since I have you on the hook right now, I'm curious to know why > the iscsi-sfnet is listing "host7" as the host. It seems like each > time I restarted the iscsi daemom this increments. That doesn't seem > normal. In 2.6 kernels the scsi layer just increments the host number every time a host is allocated. So linux-iscsi also does a host per session, so when you restart the daemon the old host is destroyed and a new one is created. > > I'll get back after testing the noop params. > for linux iscsi it is the PingTimeout, ActiveTime, and IdleTimeout. The noop params are for open-iscsi (the list you posted to :)). --~--~---------~--~----~------------~-------~--~----~ You received this message because you are subscribed to the Google Groups "open-iscsi" group. To post to this group, send email to open-iscsi@googlegroups.com To unsubscribe from this group, send email to [EMAIL PROTECTED] For more options, visit this group at http://groups.google.com/group/open-iscsi -~----------~----~----~----~------~----~------~--~---