Race condition in "take ticket"

RT Users
1

In the dark ages, I reported some problems we had with our customer
service people all clicking “take ticket” at the very same moment,
and all of them succeeding (in their own view).

This just happened to us again (twice today!) Three people clicked
on the take link at the same time, and all three of them were
assigned as the owner in succession. The last one is the only one
that “stuck”.

I’m running 3.6.3; the production box hasn’t been updated to 3.6.4
yet. Wasn’t this race squashed a while back? I’m on postgres if
that matters.

The ticket history shows it clearly:

Mon Aug 13 14:15:13 2007 pXXXXXX - Taken

Mon Aug 13 14:15:13 2007 mXXXXXX - Taken

Mon Aug 13 14:15:13 2007 sXXXXXX - Taken

2

In the dark ages, I reported some problems we had with our customer
service people all clicking “take ticket” at the very same moment,
and all of them succeeding (in their own view).

This just happened to us again (twice today!) Three people clicked
on the take link at the same time, and all three of them were
assigned as the owner in succession. The last one is the only one
that “stuck”.

I’m running 3.6.3; the production box hasn’t been updated to 3.6.4
yet. Wasn’t this race squashed a while back? I’m on postgres if
that matters.

I’m pretty sure that Todd submitted a patch that was supposed to sort
this out. Todd?

3

In the dark ages, I reported some problems we had with our customer
service people all clicking “take ticket” at the very same moment,
and all of them succeeding (in their own view).

This just happened to us again (twice today!) Three people clicked
on the take link at the same time, and all three of them were
assigned as the owner in succession. The last one is the only one
that “stuck”.

I’m running 3.6.3; the production box hasn’t been updated to 3.6.4
yet. Wasn’t this race squashed a while back? I’m on postgres if
that matters.

I’m pretty sure that Todd submitted a patch that was supposed to sort
this out. Todd?

Yeah and I’m pretty sure the fix is in the latest RT release.

4

I’m pretty sure that Todd submitted a patch that was supposed to sort
this out. Todd?

Yeah and I’m pretty sure the fix is in the latest RT release.

I thought it was sorted out too, but it just bit us after a looooong
time. I was waiting for 3.6.5 before disturbing production, but I
suppose I’ll put in 3.6.4 today and see if we can make it happen again.

5

I’m running 3.6.3; the production box hasn’t been updated to 3.6.4
yet. Wasn’t this race squashed a while back? I’m on postgres if
that matters.

FWIW, my support staff says this has been happening relatively
frequently, not just the two they told me about yesterday.

6

I’m running 3.6.3; the production box hasn’t been updated to 3.6.4
yet. Wasn’t this race squashed a while back? I’m on postgres if
that matters.

FWIW, my support staff says this has been happening relatively
frequently, not just the two they told me about yesterday.

Did you manage to get a 3.6.4 up for testing?

PGP.sig (186 Bytes)

7

I’m running 3.6.3; the production box hasn’t been updated to
3.6.4 yet. Wasn’t this race squashed a while back? I’m on
postgres if that matters.

FWIW, my support staff says this has been happening relatively
frequently, not just the two they told me about yesterday.

Did you manage to get a 3.6.4 up for testing?

i do, but it has no tickets in it, nor does it have multiple users.
I’ll do some more testing.

8

3.6.4 yet. Wasn’t this race squashed a while back? I’m on
postgres if that matters.

FWIW, my support staff says this has been happening relatively
frequently, not just the two they told me about yesterday.

Did you manage to get a 3.6.4 up for testing?

i do, but it has no tickets in it, nor does it have multiple users.
I’ll do some more testing.

Fantastic. It’s much appreciated