This post is more than 5 years old
3 Posts
0
258548
PE T320 Hyper-V Guest latency fixed by BIOS update to Host
Twice now I have had the Exchange 2010 guest on my W2K8 Standard Server with Hyper-V suddenly suffer a massive amount of latency and become practically locked up out of the blue. After hours of troubleshooting, rebooting, and researching - as a last-ditch effort I noticed that the BIOS was out of date. I upgrade BIOS and then everything magically works again.
Anywhere from 2-4 months pass and rinse & repeat.
I'm going out on a limb here but I suspect that there's something about the system that gets flushed out during a bios update that is causing the problem, but I'm at a loss to determine what that is. I want to proactively avoid having my customer's Exchange server fail, but it goes against my better judgment to apply BIOS updates proactively. We've had the motherboard swapped out already due to a bad DRAC, and the problem keeps happening.
I could use some help in identifying what else I can do to prevent this from happening in the future. TIA
dneuwir
17 Posts
0
December 17th, 2014 16:00
Well, add me to the list of (many) users seeing this issue. We're an IT MSP house, and we have about 30 of the T320/420 series units in production across various customer accounts. We have seen this identical issue (everything slows to a crawl, including the Hyper-V host), power off through the DRAC, power on, resolves for [random 1-12 weeks], then rinse and repeat.
I have even seen this on systems with the latest BIOS (as of about mid-November 2014). I have also experienced this twice on our own in-house T420.
In our experience, all of the machines with the issue have been running Hyper-V hosts--I haven't had a single-OS system with the issue yet, but that may just be coincidental. Hyper-V host ranges from Server 2008R2, to Server 2012 and Server 2012R2, so it's not any one particular Hyper-V version that causes the behavior.
I have tried seemingly everything. Every thread I've read offers a glimmer of hope. Maybe it's this, maybe it's that. Tried disabling EEE and VM Queueing on the NIC's (because evidently there's a possibly-related issue with the Broadcom's and power management that causes similar symptoms)--no avail. Tried reducing CPU cores in each VM just in case it was processor contention. No solution.
I have NOT yet tried the "Maximum Performance" setting in BIOS versus "Maximum Performance per Watt" but that will definitely be my next step. Unfortunately, because this setting involves shutting down the servers, it will be a bit before I can test. But I'm hopeful this becomes that solution I've been looking for for several years.
KeepOnSmilin
26 Posts
0
December 27th, 2014 14:00
It is coincidence. I run SBS 2011 on my T320 and every 3-4 weeks have to turn off the machine for 5 minutes, then turn it on to stop the slowness. So it is OS independent. In fact when it happens I have tried to do an Acronis restore which normally takes 10 minutes for the server, and it took over an hour. That was booting into a Linux OS.
KeepOnSmilin
26 Posts
0
December 27th, 2014 14:00
OMG!
I have been struggling with this for months. Dell tech support is unaware of the issue. I literally spent hours with them on the phone sending logs with no resolution.
I've set the bios to performance and will keep my fingers crossed.
Glad you were persistent with this LSLC. Having multiple machines certainly must have made you crazy!
I would just turn my machine off for 5 minutes and restart thought it had to do with USB or RAID.
I thank you. Dell certainly owes you big time.
jbhagan
4 Posts
0
July 17th, 2015 19:00
I think you guys may be running into this issue as well. I had similar Hyper-V performance issues. VM's would occasionally fail to respond to a ping with no network access. Otherwise in Hyper-V they looked fine. Disabling Virtual Machine Queues was the fix for me.
I realize this is an old thread but I found it today while researching the same symptoms.
AndreDuplantier
1 Message
0
September 15th, 2015 10:00
We just applied this fix to two PET320's with the exact same issue. So far so good. This thread has been a life saver (if it works). We have been messing with this issue for months and could never figure out why it randomly gets slow, then maybe a day later will go back to normal. Reboot never fixed it. Removing power did fix it.
The easiest thing to check to see if the server was in its funky state was to open Task Manager and see if the taskmgr.exe process was using around 5% CPU. If it was, then everything was slow. If it was at 0%, then everything was normal.