process dies when thread stops

[Stewart]

If start the process all threads come up w/out issue.  However, if I stop one of the threads it kills the process.  This is the error from the log

 

terminate called after throwing an instance of ‘std::logic_error’
what():  basic_string::_S_construct NULL not valid

Topic

[Jim Kosloskey]

What release of CL and does it matter which thread you stop?

email: jim.kosloskey@jim-kosloskey.com 29+ years Cloverleaf, 59 years IT - old fart.

[Stewart]

19.1.2.0

Only this process and this one thread.  Stopped / Started other threads in the process w/out issue.

 

[Rob Abbott]

Suggest you open a support case.

Rob Abbott
Cloverleaf Emeritus

[Chris]

I’m running into the same thing in our non-prd env, following an upgrade from 6.2.2 to 6.2.6. Even with verbosity turned up, there doesn’t appear to be any other related logging.

[Stewart]

I have a ticket open with cloverleaf and they couldn’t find the source of the problem either and have been unable to reproduce.  I have found that if I stop the thread, the process won’t die until a new message comes in.

[Stewart]

The process will stay up until a message comes in.  Below is the error I’m getting.  It looks like it’s trying to write a null value to the recovery DB.  If I turn off the recovery DB I’m able to stop/start the threads without issue.  I tried this with both tcpip and pdl-tcpip to rule out the PDL.

 

[pdl :PDL :DBUG/0: adt_mt:05/13/2020 08:31:52] IDLE and 4581 bytes but no error: starting READ
[pdl :PDL :DBUG/2: adt_mt:05/13/2020 08:31:52] PDL changed states: old 0, new 1
[pdl :PDL :DBUG/0: adt_mt:05/13/2020 08:31:52] Calling Tcl procedure: hci_pd.read
[pdl :PDL :DBUG/0: adt_mt:05/13/2020 08:31:52] with args: {}
[pdl :PDL :DBUG/0: adt_mt:05/13/2020 08:31:52] Tcl procedure hci_pd.read returns ‘RECEIVE’
[pdl :PDL :DBUG/0: adt_mt:05/13/2020 08:31:52] trying to match phrase: basic-msg
[pdl :PDL :DBUG/0: adt_mt:05/13/2020 08:31:52] multi_phrase_2: status = ok
[pdl :PDL :DBUG/0: adt_mt:05/13/2020 08:31:52] Calling Tcl procedure: read.done
[pdl :PDL :DBUG/0: adt_mt:05/13/2020 08:31:52] with args: {{status ok} {end {4581 0}} {data {1 4578}}}
[dbi :rlog:DBUG/3: adt_mt:05/13/2020 08:31:52] [0.0.19891] Looking for mid in recovery db
[dbi :dbi :DBUG/1: adt_mt:05/13/2020 08:31:52] (0) ‘cl_keyfindlock: About to do d_keylock’
[dbi :dbi :DBUG/1: adt_mt:05/13/2020 08:31:52] (1000) ‘cl_keyfindlock: About to do d_keyfind, keyval:’
[msg :Mid :DBUG/3: adt_mt:05/13/2020 08:31:52] Assigned mid [0.0.19891] to msg 0x7f0970018950
[msg :Msg :DBUG/0: adt_mt:05/13/2020 08:31:52] msg_Alloc new message 0x0x7f0970018950
[pdl :PDL :DBUG/0: adt_mt:05/13/2020 08:31:52] Tcl procedure read.done returns ”
[pdl :PDL :DBUG/2: adt_mt:05/13/2020 08:31:52] PDL changed states: old 1, new 0
[pdl :read:DBUG/1: adt_mt:05/13/2020 08:31:52] PDL did read msg: code = 0
[pd :pdtd:INFO/0: adt_mt:05/13/2020 08:31:52] [0.0.19885] Received msg from driver
[pd :thrd:INFO/0: adt_mt:05/13/2020 08:31:52] [0.0.19885] Placing msg on IB pre-SMS queue.
[pd :pdtd:INFO/3: adt_mt:05/13/2020 08:31:52] [0.0.19885] IB pre-SMS message details

[the msg is here]

[msg :Msg :INFO/0: adt_mt:05/13/2020 08:31:52] [0.0.19885] Writing to recovery database
[dbi :rlog:INFO/1: adt_mt:05/13/2020 08:31:52] [0.0.19885] Write msg to recovery db w/state IB pre-SMS
[dbi :log :DBUG/2: adt_mt:05/13/2020 08:31:52] log context: type 1, dbn 1, msgRec 10001, mdRec 10002, bodyRec 10003
[dbi :log :DBUG/2: adt_mt:05/13/2020 08:31:52] state 1, mode 0
terminate called after throwing an instance of ‘std::logic_error’
what(): basic_string::_S_construct NULL not valid

[Chris]

I was seeing the same, also with both tcpip and pdl-tcpip/mlp_tcp.pdl. Opened a ticket this morning, if only so Infor is aware this affects more than one customer.

[David Burks]

I think I have spoken to both people experiencing this, one by phone and other on a webex.  If anyone else encounters and creates an incident please mention my name in first event note.  I want to stay on top of this.

• David Burks

[Ramachandran R]

We also recently migrated from Cloverleaf 6.2.3 to 6.2.6, and having the same issue as mentioned in the original ticket. As a workaround we switch the thread autostart mode off, and then stop specific threads for any testing or validations.

[Tony Benitz]

We have encountered this issue as well after installing 6.2.6.0.

I have opened a support ticket.

The last message that we get in the process log before it dies is

WARNING: Message [0.0.149720326] is in the RDB and was left bound into Tcl
terminate called after throwing an instance of ‘std::logic_error’
what():  basic_string::_S_construct null not valid

Anyone find a work around or is the solution to load the 6.2.6.1 patch

 

[Stewart]

Not sure what OS version you are on but for us it is RHEL 6.5 which is no longer supported.  There are libraries missing from this version that are required.  I found that the process would stay up until a message came in and then the process died.

[Tony Benitz]

Thanks – we are on 7.8 at this time

[Tony Benitz]

It’s a BUG – Solution is HOT FIX – 6.2.6.1 per support – to bad I missed the critical fix notification before I loaded 6.2.6.0 into our production environment – now to get another downtime to load the fix.. just is life of an interface engineer I guess..

 

[Author]

Replies