Archived
This forum has been archived. Please start a new discussion on GitHub.
IceGrid/IceStorm and number of sockets limit
Dear Experts,
I have have an IceGrid deployment of IceStorm with replicated Locator.
I have four servers, each hosting IceGrid node, Locator and two instances of IceStorm.
IceGrid is configured in such a way that all clients connect to IceStorm instance #8.
I have a lot of clients and yesterday the number of open sockets on #8 hit the user limit. That particular server had very low limit.
It resulted in a weird failover to #7. Weirdness was in the fact that all nodes were in "reorganizing" state for more than a minute and apparently not able to receive or cache messages.
I will try to reproduce this behavior on my test cluster today.
Is it an expected behavior?
Thank you,
Aleksey
I have have an IceGrid deployment of IceStorm with replicated Locator.
I have four servers, each hosting IceGrid node, Locator and two instances of IceStorm.
IceGrid is configured in such a way that all clients connect to IceStorm instance #8.
I have a lot of clients and yesterday the number of open sockets on #8 hit the user limit. That particular server had very low limit.
It resulted in a weird failover to #7. Weirdness was in the fact that all nodes were in "reorganizing" state for more than a minute and apparently not able to receive or cache messages.
I will try to reproduce this behavior on my test cluster today.
Is it an expected behavior?
Thank you,
Aleksey
0