Dead Process, Log indicates ‘unable to realloc’

[Terri Scott]

I have 2 active processes on a site (1 has 5 threads and the other 6 threads).  Just about every morning, one of the processes is down.  When I check the error log, the error is always ‘unable to realloc’.  Restarting the process starts the process but don’t want to do this every morning…especially weekends.  Can anyone shed light on this error or has experienced this and was able to resolve?  I’m on 5.8. Thanks

Topic

[David Barr]

Unable to realloc means that the process has run out of memory. This can be caused by several things. You could be trying to process a message that is too large. You might also have a memory leak. You could look at any TCL procs that you have on the threads and make sure that you are releasing any memory that you allocate.

[Jim Kosloskey]

Terri,

A way to check for leaks is to stop the process and look in the log. At the end of the log the engine will indicate it is looking for leaked handles and will identify any it finds. You can go from there.

You should have the default system alerts activated which if I remember correctly will look for low memory as well as disk usage and let you know when you reach a given threshold. If you do not have those activated you might want to do that.

Depending on the O/S you are using there are tools for you to monitor the memory consumption. You might want to watch the data carefully as the day progresses to see if the memory consumption continues to rise.

Another potential is deep queues of messages I think.

email: jim.kosloskey@jim-kosloskey.com 29+ years Cloverleaf, 59 years IT - old fart.

[Doug Essner]

We were having this issue until we increased the ulimit value for memory.

[Terri Scott]

Thanks so much everyone…I will start my research based on your responses.

[Author]

Replies