gcloud can hang waiting for response from socket

[GoogleCodeExporter]

What steps will reproduce the problem?
The problem is intermittent. Just running a gcloud command that contacts the 
API often enough will eventually guarantee a hang.

What is the expected output? What do you see instead?
The expected output is the result of the command. Instead, the gcloud utility 
will hang forever and must be killed.

What is the output of 'gcloud info'?
N/A

Please provide any additional information below.

For example:
Process frozen since: Jul 17 18:20:39 server_name_here process_name: > 
/usr/local/bin/gcloud compute snapshots list some_snapshot_name_here --limit 1 
--format json

strace shows:
[user@server ~] $ sudo strace -f -p pid_here
Process ... attached
wait4(-1, 

Original issue reported on code.google.com by r...@quizlet.com on 21 Jul 2015 at 11:05

[GoogleCodeExporter]

The strace output indicates a process wait (wait4). Can you do a pstree -ap and 
post the hung gcloud command and any children? If there are children can you 
strace -f -p them?
thanks

Original comment by gsfow...@google.com on 22 Jul 2015 at 11:58

• Changed state: NeedsMoreInfo

[GoogleCodeExporter]

Hi,

I was informed by Ent. Support that this was already a known issue internally 
and to create a public issue to track it which is what I did here. I think the 
cases just still need to be linked.

Let me know if this was not the case and then next time this happens I'll get 
that additional information for you

Thanks!

Original comment by r...@quizlet.com on 22 Jul 2015 at 4:48

[GoogleCodeExporter]

There was a bit of a miscommunication here; apologies to all parties for not 
catching that sooner.

It looks like strace output from child processes is still quite needed here.

Original comment by jsche...@google.com on 11 Aug 2015 at 8:32

[GoogleCodeExporter]

I'm a coworker of ryan@quizlet, we've had another case of a gcloud command 
getting stuck. This process has at this point been stuck for over 2 days.

$ pstree -apl <GCLOUD_PID>
gcloud,<GCLOUD_PID> /usr/local/bin/gcloud compute snapshots list 
<SNAPSHOT_NAME> --limit 1 --format json
  └─python,<PY_PID> -S /usr/local/bin/../share/google/google-cloud-sdk/./lib/googlecloudsdk/gcloud/gcloud.py compute snapshots list <SNAPSHOT_NAME> --limit 1 --format json

$ strace -f -p <GCLOUD_PID>
Process 26817 attached
wait4(-1,

$strace -f -p <PY_PID>
Process 26830 attached
read(4,

Original comment by step...@quizlet.com on 13 Aug 2015 at 10:15

[GoogleCodeExporter]

I'd like to request that this be fixed with a higher priority. I don't think 
we're doing anything out of the ordinary, but both gcloud and gsutil get stuck 
for us _very frequently_. At this point they're basically unusable in any 
automated script. 

Its not acceptable for a core platform component to be this broken. Going 
forward - should we assume that gcloud and gsutil are not production ready?

Original comment by pe...@oldschoolindustriesllc.com on 14 Aug 2015 at 5:40

[GoogleCodeExporter]

Hi, This should be fixed in gcloud 0.9.74, planned for release on August 19.

Original comment by jeffvaughan@google.com on 14 Aug 2015 at 6:32

[GoogleCodeExporter]

Sounds good, thanks for your response. We'll give that a try as soon as its 
available.

Original comment by pe...@oldschoolindustriesllc.com on 14 Aug 2015 at 7:18

[GoogleCodeExporter]

Marking this as fixed. Please re-open if you're still experiencing issues with 
the latest gcloud version.

Original comment by z...@google.com on 29 Aug 2015 at 3:53

• Changed state: Fixed