command thread error

[Gary Atkinson]

I have fileset-ftp thread that grabs a charge file and sends to a outbound UPoC thread for processing the file to a particular directory and with a unique name.  Cron starts and stops the threads everday a 0800 and 0815 respectively without issues.  Inbound data matches outbound data.  But, in the process log I saw the following today:
[cmd :cmd :INFO/0:  charges_cmd:06/27/2007 08:00:39] Received command: ‘charges_xlate xrel_post’
[cmd :cmd :INFO/0:charges_xlate:06/27/2007 08:00:39] Doing ‘xrel_post’ command with args ‘‘
[cmd :cmd :INFO/0:  charges_cmd:06/27/2007 08:00:39] Inrecoverable socket error.  Closing connection.
[cmd :cmd :INFO/0:  charges_cmd:06/27/2007 08:00:39] Receiving a command
[icl :tcpi:ERR /0:  charges_cmd:06/27/2007 08:00:39] write failed: Broken pipe
[cmd :cmd :INFO/0:  charges_cmd:06/27/2007 08:00:39] Since there are some error had occured while attempted to send an Ack back to client
[cmd :cmd :INFO/0:  charges_cmd:06/27/2007 08:00:39] We try to process the command anyway but no further Ack will be send back to Client
[code]
The same group of lines repeats several times in the log.  Anyone know what causes this?[/code][code][cmd :cmd :INFO/0:  charges_cmd:06/27/2007 08:00:39] Received command: ‘charges_xlate xrel_post’
[cmd :cmd :INFO/0:charges_xlate:06/27/2007 08:00:39] Doing ‘xrel_post’ command with args ‘‘
[cmd :cmd :INFO/0:  charges_cmd:06/27/2007 08:00:39] Inrecoverable socket error.  Closing connection.
[cmd :cmd :INFO/0:  charges_cmd:06/27/2007 08:00:39] Receiving a command
[icl :tcpi:ERR /0:  charges_cmd:06/27/2007 08:00:39] write failed: Broken pipe
[cmd :cmd :INFO/0:  charges_cmd:06/27/2007 08:00:39] Since there are some error had occured while attempted to send an Ack back to client
[cmd :cmd :INFO/0:  charges_cmd:06/27/2007 08:00:39] We try to process the command anyway but no further Ack will be send back to Client
[code]
The same group of lines repeats several times in the log.  Anyone know what causes this?[/code]

Topic

[Peter Heggie]

I have this same error. It is on the inbound thread. It is a fileset-local. The input file was picked by a previous scripted step from another folder and moved to the Cloverleaf thread input folder. The script is also run by hci.

[cmd :cmd :INFO/0:  idxbill_cmd:05/01/2016 21:53:06] Received command: ‘from_idxrad pstart’

[cmd :cmd :INFO/0:  idxbill_cmd:05/01/2016 21:53:06] Doing ‘pstart’ command on ‘from_idxrad’

[prod:prod:INFO/0:  from_idxrad:05/01/2016 21:53:06] Starting protocol thread from_idxrad as tid 2.

[prod:prod:INFO/0:  from_idxrad:05/01/2016 21:53:06] Tracing status for protocol thread from_idxrad is 0.

[prod:prod:INFO/0:  from_idxrad:05/01/2016 21:53:07] Applying EO config: ”

[cmd :cmd :INFO/0:  idxbill_cmd:05/01/2016 21:53:47] Receiving a command

[cmd :cmd :INFO/0:  idxbill_cmd:05/01/2016 21:53:47] Command client went away.  Closing connection.

[cmd :cmd :INFO/0:  idxbill_cmd:05/01/2016 21:53:47] Receiving a command

[icl :tcpi:ERR /0:  idxbill_cmd:05/01/2016 21:53:47] write failed: Broken pipe

[cmd :cmd :INFO/0:  idxbill_cmd:05/01/2016 21:53:47] Since there are some error had occured while attempted to send an Ack back to client

[cmd :cmd :INFO/0:  idxbill_cmd:05/01/2016 21:53:47] We try to process the command anyway but no further Ack will be send back to Client

[cmd :cmd :INFO/0:  idxbill_cmd:05/01/2016 21:53:47] Received command: ”idxbill_xlate’

[cmd :cmd :INFO/0:  idxbill_cmd:05/01/2016 21:53:47] Cmd null in ”idxbill_xlate’

[cmd :cmd :INFO/0:  idxbill_cmd:05/01/2016 21:53:47] Inrecoverable socket error.  Closing connection.

First of all, how can you have a broken pipe error on a local file?

Could it be the internal connection from the input thread to the output thread? They are both in the same process and they are the only threads in the process. The process was started just before this thread was started.

I will try setting the threads to auto-start.

Peter Heggie

[Michael Hertel]

From what I understand this is an error from talking to the engine.

Your cron job (client) issued a command to the engine and went away (ended) instead of consuming the response (Ack) from the engine. The engine reported that its output went nowhere.

i.e.

—> hcicmd -p pm_in -c “cis_pm_in pstop”

<— Response:

<— pstop issued for thread 'cis_pm_in'

Your script ended before consuming “Response: pstop issued…”

so you see the broken pipe error which really isn’t an error.

[Peter Heggie]

great – thank you

so it was the cmd logic that was complaining. The command was received and processed but the output could not be sent.

So instead of this:

hcicmd -p $process -c “${thread} pstart” > /dev/null 2>&1

sleep 20

Should I have this?

out=`hcicmd -p $process -c “${thread} pstart”`

sleep 20

?

What is the best syntax to handle this issue for batch execution? fyi, I just changed the sleep seconds from 5 to 20. Not sure if this would help.

Peter Heggie

[Michael Hertel]

I’m not sure. I usually ignore the broken pipe errors.

I don’t use a lot of cron scripts to do things.

BTW, it looks like the error happened immediately.

So I’m not sure the sleep will help.

perhaps it’s “until” or nohup command you’re looking for?

[Michael Hertel]

Peter, here’s a post from an oldie but a goodie talking about the timeout toward the end of Rob Abbott’s explanation.

https://usspvlclovertch2.infor.com/viewtopic.php?t=364

[Peter Heggie]

thank you – that is a great explanation which completes my picture of what is going on. It also lead me to look at the documentation for hcicmd – I don’t know how long its been there, but there is a ‘-w’ parameter where the cmd timeout can be set on the invocation (so we don’t have to fool around with the perl script). The default is 30 seconds. I’m going to set this to a higher value.

Peter Heggie

[Author]

Replies