Unexpected Reboot

Bobacus

Limp Gawd
Joined
Aug 27, 2005
Messages
131
I've had a weird problem occur. Today when I went to log into my ESXi server to test Windows 7, I noticed that all of my VM's were turned off, when I had them all turned on before I logged out of the VI client last night. I checked the ESXi logs to find them empty, and when I go to the performance chart to try to stretch out the chart to find the exact time something happened, my time options are greyed out.

I've brought everything back up, and all my server OS's report unexpected shutdowns. After going through the logs, the last event's I can find range from between 8:20 and 8:40 this morning.

Just a little background for anyone curious, I'm running ESXi on a new HP ML350G5 w/ 10gb of memory, a 1tb Raid5 array on the E200i controller w/ 128mb BBUC, and Redundant PSU's. So everything is on the ESXi HCL.

The only thing I can think of is I that the VI client logon screen was open this morning when I pulled some books off my laptop keyboard, and I don't remember if I logged out of the VI client last night or not. Although even if that is the case, why would the ESXi logs be blank?
 
Well it looks like the problem was that I knocked the cord going to the UPS out of the wall this morning just enough to cause the UPS to kick over to battery, now the thing I wonder is whether the cord was loose long enough to drain the UPS battery and then I pushed it in just enough for it to regain power, or if my server did not care for the switch to battery on my Back-UPS. This is odd, since I can't reproduce the power problem, and the iLO event viewer reports power being removed and then immediately restored.

I guess I will be getting a better solution, since the right angle adaptors I have on my UPS's going to my rack arn't working too well, and I'll probably also balance the two PSU's between my battery backups.
 
I was looking at the iLo management logs, that can be accessed by logging into the onboard management web page for the server.

The power has dropped a few times this morning, and I've gotten fed up with it, since it seems that these unexpected reboots are causing issues with my virtual machines and networking. I'm in the process of installing a temp machine so I can pull all the data off my server.

I'm thinking that I will go to a straight Server 2008 install on this machine for the time being, and if I'm able to replicate the problem, contact HP and try to get it worked out. Although from what I've looked at is looks like the G5 series of servers just doesn't care for the APC Back-UPS line of battery backups.

Hopefully once I get all this worked out, I'll return to the wonders of ESXi without having to worry about my VM's becoming messed up due to power outages.
 
Wait, what? you mentioned nothing of your VMs experiencing corruptions untl the last line of the last post. I am trying to determine what would cause you to think that it's the ESXi OS, and how 2k8 will be of any use if you're experiencing power failures? Did I miss something?
 
It's actually kinda weird what has been happening. After the first or second unexpected reboot, the rdp performance on my vista virtual machine went down the tubes. I could console into it fine through the vmware infrastructure client, but RDP was unusable. One of the guys from my office and I fiddled with it for about a day or so, and I gave up with the intention to reinstall.

After this last set of reboots, my Server 2008 VM has gone to garbage when it comes to network file transfers. For example, transfers from one VM to the server VM wont go faster than 50kb/s or so.

The only thing I can think of is that on my raid config, I have it set up to use the built on drive cache, which in a power outage, any data in that cache will be lost. I've thought about turning off, but my performance goes down a bit.

The reason I am thinking of going back to 2K8 for the time being, is that I think it will be a little more tolerant to power failures than ESXi, also with going over to 2K8 I will have better access to the diagnostic tools and what not that came with the server. I'm not too concerned about it for the time being, and I don't think I'll have time to call HP till the first week of Feb when I quit my current job, and have some free time.
 
I'm going to be getting the power issues when I have time during the day to call HP, and see what they say. I'm pretty sure it's because I'm using a Back-UPS and not a Smart-UPS, but I want to hear from HP first.

The main reason I'm wanting to go back to a single OS on this server for the time being is so I have something simple that can handle the power issues. Don't get me wrong, ESXi is great, but I mainly use it for a test lab, with a storage server on it, for my stuff around the house. So I can live for awhile without my lab.

I'll let you guys know how everything goes w/ HP since I'm sure there are a few of you running the fifth generation of HP servers, that may have concerns about the power.
 
Back
Top