Incorrect message: "This server does not have sufficient memory"
Hi Folks,
We have a cluster with 3 servers (the server quota is 256MB) and suddenly the total "disk storage" and "memory cache" available was considered 0MB. The fact is that such server had about 50GB free disk space. We did remove the server from the cluster and when we tryied to add it again we got the following warning message (Completely incorrect):
"Attention - Prepare joinAttention - Prepare join failed. This server does not have sufficient memory to support the cluster quota (Cluster Quota is 256MB per node, this server only has 0MB)! failed. This server does not have sufficient memory to support the cluster quota (Cluster Quota is 256MB per node, this server only has 0MB)!"
After restarting the membase service in the problematic server, i was able to re add the server to the cluster and everything started to work as expected again. Questions:
1) Which one could be the reason that drove membase to consider that the server had not enough free space ?.
2) If this issue will happen again, is there any other option instead restarting membase service to fix the issue ?
Thanks in advance !
Alejandro
I'm not certain which build you're running, but there have been some memory accounting bugs, corner cases mostly, in older 1.6.x builds that can lead to this behavior. Restarting is the best way to fix it, I'm sorry to say. It'd be best to upgrade to 1.7.x.