Bug 240523
Summary: | Big problem with raw device binding with EMC powerpath devices, on dual core server Dell PowerEdge 2900, for Oracle RAC database server | ||
---|---|---|---|
Product: | Red Hat Enterprise Linux 4 | Reporter: | Hai Wu <hxwu> |
Component: | initscripts | Assignee: | Bill Nottingham <notting> |
Status: | CLOSED CANTFIX | QA Contact: | Brock Organ <borgan> |
Severity: | urgent | Docs Contact: | |
Priority: | medium | ||
Version: | 4.3 | CC: | andy_speagle, rick.beldin, rvokal |
Target Milestone: | --- | ||
Target Release: | --- | ||
Hardware: | i686 | ||
OS: | Linux | ||
Whiteboard: | |||
Fixed In Version: | Doc Type: | Bug Fix | |
Doc Text: | Story Points: | --- | |
Clone Of: | Environment: | ||
Last Closed: | 2008-12-08 21:15:32 UTC | Type: | --- |
Regression: | --- | Mount Type: | --- |
Documentation: | --- | CRM: | |
Verified Versions: | Category: | --- | |
oVirt Team: | --- | RHEL 7.3 requirements from Atomic Host: | |
Cloudforms Team: | --- | Target Upstream Version: | |
Embargoed: |
Description
Hai Wu
2007-05-18 05:18:51 UTC
Seems like an issue in how long it takes EMC to become available after it has started. EMC powerpath inserts 3/4 lines into file /etc/rc.sysinit, to call /etc/init.d/PowerPath start, I tried to put 5 seconds delay after that, it didn't help. I went into the script /etc/init.d/PowerPath, and add some delay, that didn't help. The only place that it helps is this /etc/sysconfig/rawdevices file. I tried to just sleep 0.2 seconds, and that also helps, I tested that for a few times, and it worked fine. It is a timing issue, but I don't know who needs to fix this. If we don't use rawdevices, EMC powerpath always works fine. It is the interoperation between these two that created this problem. It could also due to udev did not create emcpower device in time for raw command to take that as its input. Just tried to put 60 seconds delay after /etc/init.d/PowerPath start in the boot up, and that didn't help, it complained that 2 emcpower device files were not found. Is udev actually what's creating the emc devices ,or is that done by the PowerPath script? I am not sure, but in that script, it has some comments: # # Wait for udev to finish creating emcpower devices # pdfound=1 for i in 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15; do From this script, I can' tell how emc asks udev to create emcpower devices. Hm, perhaps that script should run 'udevsettle' or similar to wait for events to finish. Actually, udevsettle probably postdates RHEL 4 - never mind. I was able to make it work by adding a 15 second sleep before the "pp_start_lvm" call in the "configure_pp" function in /etc/init.d/PowerPath: --- snip rc_check sleep 15 pp_start_lvm } --- snip This seemed to allow enough time for things to settle down on my system. Of course, while this helps with my rawdevice startup bindings; I'm still having problems with LVM not finding the emcpower devices.... *grumble* ... it's definitely a timing issue that EMC needs to work on. I don't like having to hack /etc/rc.sysinit and other files to make a system wait for emcpower devices to show up. I'm going to make some modifications to the ugly /etc/init.d/PowerPath script to make this work consistently. Perhaps I can convince EMC to fix their package. Given that this seems to be solely an issue with EMC powerpath itself, I don't see anything we can fix in the initscripts themselves. Therefore, closing. (Apologies for the delay.) |