When I log on to my iDRAC via the web console it states that the version insatlled is 1.40.40, I know this is correct because i installed the latest version on each of my servers, but in OME it tells me that my servers are still using version 1.23.23.
Do you have OMSA installed along with the Inventory Collector agent (srvadmin-cm for Linux and must be installed manually). In our experience we're finding that OME doesn't show the updated version even though OMSA reports the correct version. As a result, OME continues to show that same update as still waiting to be applied and shows the server as non-compliant.
Speaking for the Linux side our findings show that the Inventory Collector agent does not run unless manually forced. While OMSA has some internal inventory mechanism, the Inventory Collector agent is another mechanism which is needed for reporting to OME. When the Inventory Collector agent does run properly, it generates the following file: /opt/dell/srvadmin/var/log/openmanage/Inventory.xml.*. When you perform a re-inventory in OME, it looks for that XML file to know what updates have been applied. Since the Inventory Collector doesn't run as expected, the XML file never gets updated and OME remains out of date.
The manual ways we force the Inventory Collector to run:
1) restart the OMSA services which will trigger the Inventory Collector to run. Grep for the process 'invcol', it will run for several minutes. Once done, it should result in an updated XML file. Then re-inventory in OME.
2) reboot the server which essentially triggers #1
I have to assume that Dell only envisioned one workflow where you always reboot after applying any package update. This doesn't make sense to me since there are plenty of updates that I'd like to apply at any time which do not require a reboot. It's rather annoying that while there's an automated way for deploying updates, I still have to manually get involved on each server if I actually want it to report back properly. There needs to be a better trigger for running the Inventory Collector.
This experience is reproducible across our entire RHEL environment. Similar results occur in our Windows environment, although some servers report back to OME correctly without restarting services or rebooting.
This is helpful information but I still see a gap in scenarios. What about applying updates via OME that do NOT require a reboot? In our testing, we experience the following behavior:
1) OME will successfully deploy the update
2) the update task completes successfully in OME with no indication of any further action or reboots required
3) we use OMSA to verify these updates were in fact installed
4) the inventory collector never gets kicked off and OME remains out of date
In this case, when should the Inventory Collector next run?
Thanks for your post. As you mentioned correctly some of the updates need reboot. That information is part of the DUP(Dell Update Package). For updates using OMSA(in-band updates) OME sends the update package to the target server and invokes the exeutable for applying that update. It also passes the flag which is part of the update wizard telling the node to reboot if required by the update package.
The check box is checked by default but if that check box is un-checked in OME and the update needs a reboot, update will not be applied till the server is rebooted either manually or by creating a reboot task in OME. If the check box is checked, OME hands the control over to DUP and it is responsible for rebooting the server after update as needed. When the server is rebooted, update is applied and inventory collector is kicked off to gather new inventory. If the update return code is successful, OME kicks off the inventory refresh cycle to get the updated component versions from the target.
For the iDrac based updates (out of band updates) OME provides the update location to iDrac and iDrac pulls the update from OME and executes the update. Some of the updates need iDrac to be rebooted in UEFI mode for the updates to be applied. iDrac has a setting called CSIOR(Collect System Inventory On Reboot) which can be set as part of iDrac configuration. If that setting is enabled, iDrac gathers the new inventory at each reboot which then can be retrieved by OME in th next inventory cycle.
Hope this gives some more clarity around how updates work and what components play a role for OME to get updated inventory information.
OME typically waits for 20 minutes or so before kicking off inventory if the success code is returned since server reboot timings can vary. You can also kick off inventory cycle from OME by right clicking on the server and selecting run discovery/inventory.
OME doesn't really kick off inventory collector run. It automatically happens at every reboot. I understand the situation now. We will have to look at the option of triggering inventory collector run remotely. We'll investigate it for future release.
Thanks for confirming. At least I now know that OME will not run the collector automatically. However, in our testing explicitly issuing a rediscover/reinventory also does not kick off the inventory collector. We tried this countless times and have waited hours and days and it just never runs. For now we just restart OMSA services to trigger the inventory. Hopefully, this can be cleaned up in a future OME update.
Inventory collector may be refershing inventory periodically without restart. If you want immidiate update, you might have to restart OMSA service if you only send updates which do not need reboot. If you have any update which needs a reboot, the OMSA service will automatically restart as part of reboot. We will inform the necessary teams so that Inventory refresh gets kicked off after updates are applied so OME can get updated information.
bobst_its
30 Posts
1828
0
Posted August 15th, 2013 10:00
Do you have OMSA installed along with the Inventory Collector agent (srvadmin-cm for Linux and must be installed manually). In our experience we're finding that OME doesn't show the updated version even though OMSA reports the correct version. As a result, OME continues to show that same update as still waiting to be applied and shows the server as non-compliant.
Speaking for the Linux side our findings show that the Inventory Collector agent does not run unless manually forced. While OMSA has some internal inventory mechanism, the Inventory Collector agent is another mechanism which is needed for reporting to OME. When the Inventory Collector agent does run properly, it generates the following file: /opt/dell/srvadmin/var/log/openmanage/Inventory.xml.*. When you perform a re-inventory in OME, it looks for that XML file to know what updates have been applied. Since the Inventory Collector doesn't run as expected, the XML file never gets updated and OME remains out of date.
The manual ways we force the Inventory Collector to run:
1) restart the OMSA services which will trigger the Inventory Collector to run. Grep for the process 'invcol', it will run for several minutes. Once done, it should result in an updated XML file. Then re-inventory in OME.
2) reboot the server which essentially triggers #1
I have to assume that Dell only envisioned one workflow where you always reboot after applying any package update. This doesn't make sense to me since there are plenty of updates that I'd like to apply at any time which do not require a reboot. It's rather annoying that while there's an automated way for deploying updates, I still have to manually get involved on each server if I actually want it to report back properly. There needs to be a better trigger for running the Inventory Collector.
This experience is reproducible across our entire RHEL environment. Similar results occur in our Windows environment, although some servers report back to OME correctly without restarting services or rebooting.
Hope this helps.