1

We have around 10 server instances (VM's) running on our Hyper-V box (mostly Windows Server 2008 R2) These do various tasks such as File server, Domain Controller, Web server, SQL Server etc.

Is there any particular regular maintenance required within these server instances?

My basic routine is :

  • Restart each VM (Monthly)
  • Remove any Temp files / general disk cleanup (Monthly)
  • Check all windows updates are applied (Weekly)
  • Check error logs for anything unusual (Weekly)
  • Check Task manager for anything odd (Weekly)

Anything else I really should be doing? Is de-fragmentation necessary?

Digital Lightcraft
  • 369
  • 4
  • 19
  • You might want to clarify you are asking for recommended maintenance on the Hyper-V GUESTS, whereas saying "hyper-v servers" implies maintenance on the Hyper-V HOST. – jlehtinen Jan 09 '14 at 16:26
  • ok, i thought that was fairly obvious but i will! – Digital Lightcraft Jan 09 '14 at 16:40
  • 1
    Actually no. A guest is a normal server - nothing hyper-v specific here. A hyper-V server is the host. – TomTom Jan 09 '14 at 16:50
  • 1
    I would suggest weekly oiling of all rotating parts, specifically the GDI and the Powershell module. Also keep in mind that the .NET framework would require frequent disassembly for proper garbage collection. – the-wabbit Jan 09 '14 at 17:08
  • 1
    Especially because some of the parts of the .NET framework are already old and quite rusty. Require regular oiling too - and definitely sometimes brushing so they look shiny. – TomTom Jan 09 '14 at 17:08
  • Seriously: you should have monitoring in place and an operations guide and a security concept written for the services you're running. There is nothing OS-specific which needs to be done on a regular basis, except maybe from security patches - but frequency and timelines would depend on what you have defined in the aforementioned documents. – the-wabbit Jan 09 '14 at 17:12

1 Answers1

3

•Restart each VM (Monthly) - This is debatable. I don't reboot servers unless it's needed to resolve an issue, apply an update or make a change that requires a reboot.

•Remove any Temp files / general disk cleanup (Monthly) - IMO, this is a pointless pursuit.

•Check all windows updates are applied (Weekly) - You should be installing Windows updates on a regular schedule. The actual scheduling of installing updates is completely up to you.

•Check error logs for anything unusual (Weekly) - You should be using automation to monitor your event logs and using email alerts based on those event log entries (errors, warnings, critical)

•Check Tack manager for anything odd (Weekly) - You should be using automation to monitor resource usage on your VM's and using email alerts to notify you of resource usage that falls outside of your defined parameters.

joeqwerty
  • 109,901
  • 6
  • 81
  • 172
  • Not really on updates. MS publishes updates normally once per month on second tuesday. There is normally no sense in doing weekly updates. If one is smart enough to keep an eye open for the WIDELY published emergency issues that may come up once or twice per year. 2nd Tuesday of a month is patch release day. – TomTom Jan 09 '14 at 17:08
  • @joeqwerty - when you have limited disk space, clearing up the crap that windows leaves behind is quite necessary - I freed up a good 3GB between the VM's in one day recently. Automation cant spot everything - Last month the web server was lagging - there was a strange high CPU load for 1 minute every 2 minutes, this couldnt have been spotted by automation (solved by a restart too, never did figure out what it was, but its not happened again, yet) – Digital Lightcraft Jan 09 '14 at 17:18
  • @DarkcatStudios - No offense but if you have to clean out temp files to keep your VM's disks from running out of space then you're doing it wrong. Expand the virtual disks for the affected VM's. If you don't have enough physical storage to expand the virtual disks then you're doing it wrong. If you're employer/client won't allocate the funds needed to implement a proper storage infrastructure then they're doing it wrong. In addition, stating that automation can't help you solve your problems is a dumb-headed approach to take. You certainly could have caught the issue with proper monitoring. – joeqwerty Jan 09 '14 at 17:31
  • 2
    Are you saying that automation wouldn't have caught the problem but that you'll catch it by manually checking Task Manager a few times a day or week? You're in for a very arduous, exhausting career using your current approach to system administration. Again, I don't mean any offense but you need to work smart, not hard. – joeqwerty Jan 09 '14 at 17:33
  • you are clearly "up yourself" - Starting a comment with "no offence" does not make in un-offensive... The sort of activity I was seeing could not be caught by monitoring. – Digital Lightcraft Jan 13 '14 at 09:34
  • "you are clearly up yourself" - What does that mean? – joeqwerty Jan 13 '14 at 16:52
  • @darkcat actually monitoring would have caught such a thing. By monitoring page load times for the webserver. And showing you a graph of cpu usage. Automated monitoring does what you can't no matter how hard you work - watch every server every minute of every day. And when you find something it didn't catch you reconfigure the monitoring system to catch it next time. – Grant Apr 03 '14 at 23:57
  • 1
    @Grant - What sort of monitoring tool(s) would you recommend? – Digital Lightcraft Apr 10 '14 at 08:02