This post is more than 5 years old
2 Intern
•
178 Posts
0
2192
VNXe memory leak bug in code / Monitoring cache limit in unisphere
Recently We have experienced memory leak error in ECOM (unisphere) which caused sotrage servers connected to SPB restart.
We were informed that it's bug in current code we are using and suggested to perform the upgrade. Meanwhile,
Is there any way that we can monitor the cache limit on unisphere and restart management servers once cache threshold is reached.
Thanks
Shardul1
33 Posts
0
October 29th, 2013 00:00
Hello,
mgmtd memory leak has been a very common culprit for causing SP reboots on codes before MR3SP1.1.
You are right. Restarting mgmtd process at the right time does help us avoid the reboot. On code MR3SP1.1, the implemented workaround actually restarts mgmtd process if it consumes more than 1GB of memory.
However, code MR4 provides a more definitive solution to the problem. Upgrading to MR4 avoids the condition in which mgmtd would consume so much memory (prevents the memory leak).
Regards,
Shardul
khkris
2 Intern
2 Intern
•
178 Posts
0
October 29th, 2013 08:00
Thanks Shardul.
Any suggestions on how can we verify consumed sapce to restart when ever it reaches more than 1 GB?
How much cache memory is allocated to Unisphere by default?
Thanks
Shardul1
33 Posts
1
October 29th, 2013 08:00
Hello,
The memory consumed by mgmtd process can be monitored by running 'top' command on the master SP. Max memory available is 2GB. Beyond that 1GB swap is also available. The panic occurs when all this memory is exhausted.
Regards