Network workstations freeze for a few seconds every few days
Every few days, though not at any specific time, users say that their application freezes for 5-10 seconds. The applications are typically Word or InDesign. I have a user who texts me every time this happens so I have a record of the time.
Server 2003 peer-to-peer setup. Workstations are Windows 7. Server just has files and a network printer, but also runs Hamachi for a handful of remote users to gain access. Runs DHCP and DNS. No email or SQL. Nothing in any of the Event Logs. Nothing like cloud backup, local backup, shadow copy, remote access seems to coincide with the problem.
It's an old server. Dell PowerEdge 830, Pentium D 3 GHz, 2 GB RAM, CERC SATA 1.5 RAID controller with two pairs of mirrored drives (so 4 drives total). OS on one set, user data on the other.
I'm seeing instances where the Avg Disk Queue Length is at 100 for minutes at a time, but again, no pattern that I can see.
Consumer-grade 100 Mb 16-port unmanaged switch. Most workstations are wired. There are a few 5-port switches scattered about the office because they don't have enough drops, but everyone on the network, regardless of where they're located in the office, will notice a freeze.
Any thoughts on how I can isolate the problem? Is the server just tapped out?
Server 2003 peer-to-peer setup. Workstations are Windows 7. Server just has files and a network printer, but also runs Hamachi for a handful of remote users to gain access. Runs DHCP and DNS. No email or SQL. Nothing in any of the Event Logs. Nothing like cloud backup, local backup, shadow copy, remote access seems to coincide with the problem.
It's an old server. Dell PowerEdge 830, Pentium D 3 GHz, 2 GB RAM, CERC SATA 1.5 RAID controller with two pairs of mirrored drives (so 4 drives total). OS on one set, user data on the other.
I'm seeing instances where the Avg Disk Queue Length is at 100 for minutes at a time, but again, no pattern that I can see.
Consumer-grade 100 Mb 16-port unmanaged switch. Most workstations are wired. There are a few 5-port switches scattered about the office because they don't have enough drops, but everyone on the network, regardless of where they're located in the office, will notice a freeze.
Any thoughts on how I can isolate the problem? Is the server just tapped out?
ASKER
Thanks for your response.
Since I'm not on site and the freeze lasts only a few minutes, I can't get to the server fast enough to see if the high disk queues are also when the users are experiencing problems. I run LogMeIn Central and have looked at the alerts, but they don't include disk queue.
I'm already running Performance Monitor on the server and check on it a few times a day. I see spikes in RAM, processor and disk queues, then occasionally the disk queues will be pegged at 100 with users just having a few files open.
But those files are often InDesign and the file sizes are always > 20 MB. Maybe the disk queues are peaking when they read/write from these InDesign files. I haven't been able to confirm that suspicion.
I'll set up a workstation with a constant ping and see it loses connectivity.
Since I'm not on site and the freeze lasts only a few minutes, I can't get to the server fast enough to see if the high disk queues are also when the users are experiencing problems. I run LogMeIn Central and have looked at the alerts, but they don't include disk queue.
I'm already running Performance Monitor on the server and check on it a few times a day. I see spikes in RAM, processor and disk queues, then occasionally the disk queues will be pegged at 100 with users just having a few files open.
But those files are often InDesign and the file sizes are always > 20 MB. Maybe the disk queues are peaking when they read/write from these InDesign files. I haven't been able to confirm that suspicion.
I'll set up a workstation with a constant ping and see it loses connectivity.
ASKER CERTIFIED SOLUTION
membership
This solution is only available to members.
To access this solution, you must be a member of Experts Exchange.
ASKER
The server is tapped out. It's old and time to replace it. Thanks for your help regarding using PerfMon logging.
anyway to reduce the guesswork you need to fid out narrow it down a bit.
run performance monitors on the server and also leave some pings running from some PCs.
when the issue occurs you can check if the pings are OK. if yes its not network related.
also check the server perf logs to see if any spikes in disk/network/cpu/memory occur around the same time.