jonathanmetzman

Can you run the formatter please.

/gcbrun

Fixes: https://pantheon.corp.google.com/errors/detail/CPquw-3h_ueTvQE;service=;version=?project=google.com:cluster-fuzz

Two that have been doing their tasks for weeks are stuck here: `Testing for crash`

I've looked at two instances suffering from this issue. Their stacktraces looked roughly similar: ``` (gdb) bt #0 0x00005cbe7d5d4b34 in sre_ucs1_match (state=state@entry=0x7ffee26f5bd0, pattern=pattern@entry=0x5cbe7fc7c014, toplevel=toplevel@entry=0) at ./Modules/sre_lib.h:590 #1 0x00005cbe7d5db30d in sre_ucs1_search...

Got a stacktrace from a third: ``` Traceback (most recent call last): File "/mnt/scratch0/clusterfuzz/src/python/bot/startup/run_bot.py", line 249, in main() File "/mnt/scratch0/clusterfuzz/src/python/bot/startup/run_bot.py", line 212, in main error_stacktrace, clean_exit, task_payload = task_loop() File...

I think a few hundred bots are blocked by this.

Another thing we should have is better task killing. A bot should kill a process that runs a task for too long. @oliverchang Do you know if we have this...

Another lesson we can learn is to have alerts when queue backlogs get too long. This is both a symptom of problems and an issue in of itself.

This has been mitigated in ClusterFuzz and fuzztest no longer does this.