Jump to content


TCP Send Failed - cannot submit large changelists

submittcp error WSAECONNABORTED

  • Please log in to reply
1 reply to this topic

#1 ironbelly

ironbelly

    Newbie

  • Members
  • Pip
  • 9 posts

Posted 13 March 2017 - 06:14 PM

We are constantly being disconnected in the middle of large commits and I can't figure out why so I come here with my hands open hoping some kind soul can help me diagnose and cure us of this problem.    It happens with submits of different shapes or sizes but in the end for submits of a certain size and above it always fails.  When running this from CLI the error message I receive is:

Quote


TCP send failed.
write: socket: WSAECONNABORTED
Perforce client error:
        TCP send failed.
        write: socket: WSAECONNABORTED

Interestingly enough when I run this very same changelist in CLI without parallel threads enabled it seems to fail sooner(20-30 seconds) , and this


In P4V it just says: Submit aborted -- fix problems then use 'p4 submit -c 208'

When I look on the server logs I see:

Quote



Perforce server info:
        2017/03/13 10:34:31 pid 7927 completed 808s 2978+437us 0+46672io 0+0net 5664k 0pf
Perforce server info:
        2017/03/13 10:21:02 pid 7927 IB_Admin@IB_Admin_DESKTOP-H57FG1F_FRV_2017_1663 xx.xx.xx.xx [p4/2015.2/NTX64/1443805] 'user-transmit -t 7919 -b 8 -r'
--- lapse 808s
--- usage 2978+437us 0+46672io 0+0net 5664k 0pf
--- rpc msgs/size in+out 9255+19/36mb+0mb himarks 318788/523588 snd/rcv .000s/802s
--- rpc receive errors, duplexing F/R 1075/1

Perforce server error:
        Date 2017/03/13 10:34:31:
        Pid 7927
        Connection from xx.xx.xx.xx:54385 broken.
        RpcTransport: partial message read
        TCP receive failed.
        read: socket: Connection timed out

I'm not necessarily saying this is an issue with the perforce server, it could be with or connection, it could be with our router, but I was hoping someone might have some ideas that could point me in the right direction.  Thanks!

#2 p4rfong

p4rfong

    Advanced Member

  • Staff Moderators
  • 343 posts

Posted 17 March 2017 - 01:42 AM

First turn off parallel submits for now.  Then try a different TCP size and filesys.bufsize variables to see if any helps.

p4 -v net.tcpsize=64K submit -c <changelist>

p4 -v net.tcpsize=2M submit -c <changelist>

p4 -v filesys.bufsize=2M submit -c <changelist>

p4 -v filesys.bufsize=2K submit -c <changelist>





Also tagged with one or more of these keywords: submittcp, error, WSAECONNABORTED

0 user(s) are reading this topic

0 members, 0 guests, 0 anonymous users