RTXteam
kg2 problem query?
I was trying to do some testing at /kg2NewFmt and it seemed like the connection to Apache was in limbo. I sent in the query, but the connection was stuck. Not rejected, not failed, just hanging.
Looking on the server, I saw two big bloated processes using 100% CPU and 8 GB RAM. I just manually killed them both. This almost immediately unhung my issues query to /kg2NewFmt and they completed rapidly. As far as I can tell, the queries that I killed were to production kg2. It is totally unclear to me why an ailing production kg2 query would cause connection hanging/deferment at NewFmt kg2. weird. They should be completely separate processes.
I then discovered that production kg2 was down. Because I killed it. duh. I'm generally used to killing child processes not parent processes. But I suppose this is expected. kg2 does not spawn children.
I restarted kg2 and it seems okay. But I don't fully understand what happened.
The query in question is one of these two I suspect:
I suspect the second one was the killer So this query: https://arax.test.transltr.io/api/arax/v1.3/status?id=881542 may be a kg2 killer. perhaps looking into if someone has time. But carefuly, it may cause kg2 to spin out of control.
There were some other problems in the same time frame. Unclear if they are directly related:
