Hundreds of XAPI processes
-
Hi there!
I have an issue with my homelab 8.2.1 host. This is a very modest machine; an i7-3770k with 32G of RAM, 4T of RAID10 and a 128G RAID1 boot disk.
My machine occasionally (perhaps every 3-4 months) will go crazy with hundreds of spawned /opt/xensource/bin/xapi processes. They have chewed up all the RAM on dom0. The box is crawling because it's in swap now.
Is there a way to safely restart the xapi processes without a full host reboot?
(And I have a confession to make: because I have so little RAM...32G is the max for this chipset...I have tweaked the RAM on dom0 down to 1.5G. Again, home lab....not production.)
-
Hmm. Also hundreds of /usr/sbin/CROND processes running "/usr/lib64/sa/sadc -F _L -S DISK 1 1 -".
Something is up with this box.
-
Do a toolstack restart, it's safe for your running VMs
It might be interesting to crawl the logs to understand what's going on.
-
@olivierlambert I tried restarting the toolstack, but it never came back. I killed all the cron processes, leaving hundreds of the associated dead "sadc" processes .
Not wanting to end up breaking something with the big hammer, I ended up simply having to reboot. That box runs my main workstation and my firewall for the house, so the natives were getting restless. Everything came back up just fine.
Not sure what happened there. The first indication that something was wrong was that the Marvell RAID software I installed for my cheap RAID controller (as seen in this thread) had spawned hundreds of apache processes.
At this point, I'm going to chalk it up to the box simply running out of RAM, and guessing that it was that RAID card software.
I appreciate you guys being available in-person in the support forums for freeloaders like me.
-
You could maybe raise the memory a bit, but yeah, the software RAID thing might hurt more than anything else
Maybe switch to full soft RAID with mdadm, that might be better
