• Introducing Dark Mode! Switch by clicking on the lightbulb icon next to Search or by clicking on Default style at the bottom left of the page!

Semaphore/Mutex Problem

mats_thunholm

Active Member
Dear JDEList,

We are running 'OW B7332, SP10, AS400 as an ES'.
Lately, we have had 'Semaphore/Mutex' problems between some of our
'subsystem jobs' (UBEs).
It seems that more than one subsystem job wants the same semaphore value.
I.e. we kick off R42565/ETH20006, after 5 min we kick off R47011/ETH0003.
Job R47011/ETH0003 goes into a status of SEMW (Semaphore Wait) or sometimes
MTXW (Mutex Wait),
the only way to get R47011 thru is to delete the other job (as they are
'subsystem jobs' they are always active and
keep there semaphore value for ever, I guess). Usually when we kill the
other job the specs are corrupted and we
have to recreate our specs again otherwise no UBE will work.
Is there something that can be done to solve this ??

Thanks
Mats Thunholm

---------------------------------------------------------------------------------------------------------

Mats Thunholm
APBiotech
Uppsala, Sweden
mats.thunholm@eu.apbiotech.com
---------------------------------------------------------------------------------------------------------



B7332 SP10, AS400 V4R5
Mats Thunholm
APBiotech
Uppsala, Sweden
 
Hi,
When you set the jobd to be multithread, did you set the jobq maximum job to
be more than 1 as well. we've encountered this before, which is only 1
report is running at a time for some time and the other jobs are in mutex
status. but after we've changed the maximum jobs to be e.g. 5, then the
reports will run concurrently. maybe you should check your setting again.
here's how to do it:
1) make sure your job description allow multithread is set to *yes
2) change your jobq entry maximum active jobs to a value you want to run
concurrent(e.g. 5), instead of using the *same value
3) go to dspsbsd and check your jobq entry. you'll notice that the jobq in
this sub-system will have an "max active" which is not 1.
try to run your reports again, concurrently to this sub-system and you can
notice that all job you're running will be concurrent. in jde ow "wsj", you
can see all the job you're running is in "procesing" status.

Cheers,
Bell
PS this assumes you've already had your setting inline, which is your jobq
is already tied to a sub-system, and the only extra setting here is to
change your jobq and jobd config.

----- Original Message -----
From: "mats thunholm" <mats.thunholm@eu.apbiotech.com>
To: <jdelistml@jdelist.com>
Sent: Sunday, February 25, 2001 9:37 PM
Subject: Semaphore/Mutex Problem


> Dear JDEList,
>
> We are running 'OW B7332, SP10, AS400 as an ES'.
> Lately, we have had 'Semaphore/Mutex' problems between some of our
> 'subsystem jobs' (UBEs).
> It seems that more than one subsystem job wants the same semaphore value.
> I.e. we kick off R42565/ETH20006, after 5 min we kick off R47011/ETH0003.
> Job R47011/ETH0003 goes into a status of SEMW (Semaphore Wait) or
sometimes
> MTXW (Mutex Wait),
> the only way to get R47011 thru is to delete the other job (as they are
> 'subsystem jobs' they are always active and
> keep there semaphore value for ever, I guess). Usually when we kill the
> other job the specs are corrupted and we
> have to recreate our specs again otherwise no UBE will work.
> Is there something that can be done to solve this ??
>
> Thanks
> Mats Thunholm
>
> --------------------------------------------------------------------------
-------------------------------
>
> Mats Thunholm
> APBiotech
> Uppsala, Sweden
> mats.thunholm@eu.apbiotech.com
> --------------------------------------------------------------------------
-------------------------------
>
>
>
> B7332 SP10, AS400 V4R5
> Mats Thunholm
> APBiotech
> Uppsala, Sweden
>
> --------------------------
> To view this thread, visit the JDEList forum at:
>
http://198.144.193.139/cgi-bin/wwwthreads/showflat.pl?Cat=0&Board=OW&Number=
6226
> *************************************************************
> This is the JDEList One World / XE Mailing List.
> Archives and information on how to SUBSCRIBE, and
> UNSUBSCRIBE can be found at http://www.JDELIST.com
> *************************************************************
>
 

mats_thunholm

Active Member
We installed that SAR (3696736) 2000-08-22.

Are maybe my CPI values in the Server INI file to low ?

[JDEIPC]
maxNumberOfResources=1000
startIPCKeyValue=2101
avgResourceNameLength=15
maxMsgqEntries=1024
maxMsgqBytes=65536
ipcTrace=0

Thanks
Mats Thunholm

---------------------------------------------------------------------------------------------------------

Mats Thunholm
APBiotech
Uppsala, Sweden
mats.thunholm@eu.apbiotech.com
---------------------------------------------------------------------------------------------------------



"JDNowell"
<hotm6654@hotmail To: <jdelist@jdelist.com>, <jdelistml@jdelist.com>
.com> cc:
Sent by: Subject: Re: Semaphore/Mutex Problem
owner-jdelistml@j
delist.com


2001-02-26 22:09
Please respond to
jdelist






Checkout SAR 3696736 / ESU JD4618. That should take care of your problem.


----- Original Message -----
From: "belladandy" <belladandy@mailandnews.com>
To: <jdelistml@jdelist.com>
Sent: Sunday, February 25, 2001 9:11 PM
Subject: Re: Semaphore/Mutex Problem


> Hi,
> When you set the jobd to be multithread, did you set the jobq maximum job
to
> be more than 1 as well. we've encountered this before, which is only 1
> report is running at a time for some time and the other jobs are in mutex
> status. but after we've changed the maximum jobs to be e.g. 5, then the
> reports will run concurrently. maybe you should check your setting again.
> here's how to do it:
> 1) make sure your job description allow multithread is set to *yes
> 2) change your jobq entry maximum active jobs to a value you want to run
> concurrent(e.g. 5), instead of using the *same value
> 3) go to dspsbsd and check your jobq entry. you'll notice that the jobq
in
> this sub-system will have an "max active" which is not 1.
> try to run your reports again, concurrently to this sub-system and you
can
> notice that all job you're running will be concurrent. in jde ow "wsj",
you
> can see all the job you're running is in "procesing" status.
>
> Cheers,
> Bell
> PS this assumes you've already had your setting inline, which is your
jobq
> is already tied to a sub-system, and the only extra setting here is to
> change your jobq and jobd config.
>
> ----- Original Message -----
> From: "mats thunholm" <mats.thunholm@eu.apbiotech.com>
> To: <jdelistml@jdelist.com>
> Sent: Sunday, February 25, 2001 9:37 PM
> Subject: Semaphore/Mutex Problem
>
>
> > Dear JDEList,
> >
> > We are running 'OW B7332, SP10, AS400 as an ES'.
> > Lately, we have had 'Semaphore/Mutex' problems between some of our
> > 'subsystem jobs' (UBEs).
> > It seems that more than one subsystem job wants the same semaphore
value.
> > I.e. we kick off R42565/ETH20006, after 5 min we kick off
R47011/ETH0003.
> > Job R47011/ETH0003 goes into a status of SEMW (Semaphore Wait) or
> sometimes
> > MTXW (Mutex Wait),
> > the only way to get R47011 thru is to delete the other job (as they are
> > 'subsystem jobs' they are always active and
> > keep there semaphore value for ever, I guess). Usually when we kill the
> > other job the specs are corrupted and we
> > have to recreate our specs again otherwise no UBE will work.
> > Is there something that can be done to solve this ??
> >
> > Thanks
> > Mats Thunholm
> >
>
>
--------------------------------------------------------------------------
> -------------------------------
> >
> > Mats Thunholm
> > APBiotech
> > Uppsala, Sweden
> > mats.thunholm@eu.apbiotech.com
>
>
--------------------------------------------------------------------------
> -------------------------------
> >
> >
> >
> > B7332 SP10, AS400 V4R5
> > Mats Thunholm
> > APBiotech
> > Uppsala, Sweden
> >
> > --------------------------
> > To view this thread, visit the JDEList forum at:
> >
>
http://198.144.193.139/cgi-bin/wwwthreads/showflat.pl?Cat=0&Board=OW&Number
=
> 6226
> > *************************************************************
> > This is the JDEList One World / XE Mailing List.
> > Archives and information on how to SUBSCRIBE, and
> > UNSUBSCRIBE can be found at http://www.JDELIST.com
> > *************************************************************
> >
>
>
>
>
>
> --------------------------
> To view this thread, visit the JDEList forum at:
>
http://198.144.193.139/cgi-bin/wwwthreads/showflat.pl?Cat=0&Board=OW&Number
=
6235
> *************************************************************
> This is the JDEList One World / XE Mailing List.
> Archives and information on how to SUBSCRIBE, and
> UNSUBSCRIBE can be found at http://www.JDELIST.com
> *************************************************************
>
>
*************************************************************
This is the JDEList One World / XE Mailing List.
Archives and information on how to SUBSCRIBE, and
UNSUBSCRIBE can be found at http://www.JDELIST.com
*************************************************************






B7332 SP10, AS400 V4R5
Mats Thunholm
APBiotech
Uppsala, Sweden
 

hotm6654

Well Known Member
How many JDENET_N kernels are running? How many UBE JDENET_K kernels (Type
2) are you running?

This surprises me that you would have 3696736.



----- Original Message -----
From: "mats thunholm" <mats.thunholm@eu.apbiotech.com>
To: <jdelistml@jdelist.com>
Sent: Tuesday, February 27, 2001 6:01 AM
Subject: Re: Semaphore/Mutex Problem


>
> We installed that SAR (3696736) 2000-08-22.
>
> Are maybe my CPI values in the Server INI file to low ?
>
> [JDEIPC]
> maxNumberOfResources=1000
> startIPCKeyValue=2101
> avgResourceNameLength=15
> maxMsgqEntries=1024
> maxMsgqBytes=65536
> ipcTrace=0
>
> Thanks
> Mats Thunholm
>
> --------------------------------------------------------------------------
-------------------------------
>
> Mats Thunholm
> APBiotech
> Uppsala, Sweden
> mats.thunholm@eu.apbiotech.com
> --------------------------------------------------------------------------
-------------------------------
>
>
>
> "JDNowell"
> <hotm6654@hotmail To:
<jdelist@jdelist.com>, <jdelistml@jdelist.com>
> .com> cc:
> Sent by: Subject: Re:
Semaphore/Mutex Problem
> owner-jdelistml@j
> delist.com
>
>
> 2001-02-26 22:09
> Please respond to
> jdelist
>
>
>
>
>
>
> Checkout SAR 3696736 / ESU JD4618. That should take care of your problem.
>
>
> ----- Original Message -----
> From: "belladandy" <belladandy@mailandnews.com>
> To: <jdelistml@jdelist.com>
> Sent: Sunday, February 25, 2001 9:11 PM
> Subject: Re: Semaphore/Mutex Problem
>
>
> > Hi,
> > When you set the jobd to be multithread, did you set the jobq maximum
job
> to
> > be more than 1 as well. we've encountered this before, which is only 1
> > report is running at a time for some time and the other jobs are in
mutex
> > status. but after we've changed the maximum jobs to be e.g. 5, then the
> > reports will run concurrently. maybe you should check your setting
again.
> > here's how to do it:
> > 1) make sure your job description allow multithread is set to *yes
> > 2) change your jobq entry maximum active jobs to a value you want to run
> > concurrent(e.g. 5), instead of using the *same value
> > 3) go to dspsbsd and check your jobq entry. you'll notice that the jobq
> in
> > this sub-system will have an "max active" which is not 1.
> > try to run your reports again, concurrently to this sub-system and you
> can
> > notice that all job you're running will be concurrent. in jde ow "wsj",
> you
> > can see all the job you're running is in "procesing" status.
> >
> > Cheers,
> > Bell
> > PS this assumes you've already had your setting inline, which is your
> jobq
> > is already tied to a sub-system, and the only extra setting here is to
> > change your jobq and jobd config.
> >
> > ----- Original Message -----
> > From: "mats thunholm" <mats.thunholm@eu.apbiotech.com>
> > To: <jdelistml@jdelist.com>
> > Sent: Sunday, February 25, 2001 9:37 PM
> > Subject: Semaphore/Mutex Problem
> >
> >
> > > Dear JDEList,
> > >
> > > We are running 'OW B7332, SP10, AS400 as an ES'.
> > > Lately, we have had 'Semaphore/Mutex' problems between some of our
> > > 'subsystem jobs' (UBEs).
> > > It seems that more than one subsystem job wants the same semaphore
> value.
> > > I.e. we kick off R42565/ETH20006, after 5 min we kick off
> R47011/ETH0003.
> > > Job R47011/ETH0003 goes into a status of SEMW (Semaphore Wait) or
> > sometimes
> > > MTXW (Mutex Wait),
> > > the only way to get R47011 thru is to delete the other job (as they
are
> > > 'subsystem jobs' they are always active and
> > > keep there semaphore value for ever, I guess). Usually when we kill
the
> > > other job the specs are corrupted and we
> > > have to recreate our specs again otherwise no UBE will work.
> > > Is there something that can be done to solve this ??
> > >
> > > Thanks
> > > Mats Thunholm
> > >
> >
> >
> --------------------------------------------------------------------------
> > -------------------------------
> > >
> > > Mats Thunholm
> > > APBiotech
> > > Uppsala, Sweden
> > > mats.thunholm@eu.apbiotech.com
> >
> >
> --------------------------------------------------------------------------
> > -------------------------------
> > >
> > >
> > >
> > > B7332 SP10, AS400 V4R5
> > > Mats Thunholm
> > > APBiotech
> > > Uppsala, Sweden
> > >
> > > --------------------------
> > > To view this thread, visit the JDEList forum at:
> > >
> >
>
http://198.144.193.139/cgi-bin/wwwthreads/showflat.pl?Cat=0&Board=OW&Number
> =
> > 6226
> > > *************************************************************
> > > This is the JDEList One World / XE Mailing List.
> > > Archives and information on how to SUBSCRIBE, and
> > > UNSUBSCRIBE can be found at http://www.JDELIST.com
> > > *************************************************************
> > >
> >
> >
> >
> >
> >
> > --------------------------
> > To view this thread, visit the JDEList forum at:
> >
>
http://198.144.193.139/cgi-bin/wwwthreads/showflat.pl?Cat=0&Board=OW&Number
> =
> 6235
> > *************************************************************
> > This is the JDEList One World / XE Mailing List.
> > Archives and information on how to SUBSCRIBE, and
> > UNSUBSCRIBE can be found at http://www.JDELIST.com
> > *************************************************************
> >
> >
> *************************************************************
> This is the JDEList One World / XE Mailing List.
> Archives and information on how to SUBSCRIBE, and
> UNSUBSCRIBE can be found at http://www.JDELIST.com
> *************************************************************
>
>
>
>
>
>
> B7332 SP10, AS400 V4R5
> Mats Thunholm
> APBiotech
> Uppsala, Sweden
>
> --------------------------
> To view this thread, visit the JDEList forum at:
>
http://198.144.193.139/cgi-bin/wwwthreads/showflat.pl?Cat=0&Board=OW&Number=
6264
> *************************************************************
> This is the JDEList One World / XE Mailing List.
> Archives and information on how to SUBSCRIBE, and
> UNSUBSCRIBE can be found at http://www.JDELIST.com
> *************************************************************
>
>



JD Nowell
OW: B7332
ES: AS400 V4R4 CO: DB2/400 SP: 11.2
Users: 250 TSE Users: 100
 

mats_thunholm

Active Member
Hi List,

We are running 1 JDENET_N kernel
We are running 1-2 JDENET_K kernels (def2, ube)
We are running 4-8 JDENET_K Kernels (def12, ubesbs)

the JDENET_K (def12, ubesbs) jobs are increasing if our 'subsystem ube's'
are
bumping out due to SEMW/MTXW. Usually when we start up in the morning we
have about
2-4 JDENET_K (def12, ubesbs) but as long as the sbs jobs are crashing out
that amount is increasing.

see also our INI setting.

[JDENET]
netPgmName=/QSYS.LIB/B7332SYS.LIB/JDENET_N.PGM
krnlPgmName=/QSYS.LIB/B7332SYS.LIB/JDENET_K.PGM
serviceNameListen=6008
serviceNameConnect=6008
maxNetProcesses=1
maxNetConnections=800
newProcessThresholdConnects=0
netShutdownInterval=15
connectTimeout=10
maxKernelProcesses=30
maxKernelRanges=12
kernelDelay=0
maxLenInlineData=1024
maxLenFixedData=4096
maxFixedDataPackets=1024
singleProcess=0
netTrace=0
enableMsgPriority=0
enableGDMsgs=0
gdmRetryInterval=60
krnlCoreDump=0

[JDENET_KERNEL_DEF2]
dispatchDLLName=JDEKRNL
dispatchDLLFunction=JDEK_DispatchUBEMessage
maxNumberOfProcesses=10
beginningMsgTypeRange=256
endingMsgTypeRange=511
newProcessThresholdRequests=1
numberOfAutoStartProcesses=0

[JDENET_KERNEL_DEF12]
dispatchDLLName=JDEKRNL
dispatchDLLFunction=JDEK_DispatchUBESBSMessage
maxNumberOfProcesses=1
beginningMsgTypeRange=3501
endingMsgTypeRange=3756
newProcessThresholdRequests=1
numberOfAutoStartProcesses=0


Thanks
Mats Thunholm

---------------------------------------------------------------------------------------------------------

Mats Thunholm
APBiotech
Uppsala, Sweden
mats.thunholm@eu.apbiotech.com
---------------------------------------------------------------------------------------------------------



hotm6654
<hotm6654@hotmail To: jdelistml@jdelist.com
.com> cc:
Sent by: Subject: Re: Semaphore/Mutex Problem
owner-jdelistml@j
delist.com


2001-02-27 15:28
Please respond to
jdelist






How many JDENET_N kernels are running? How many UBE JDENET_K kernels (Type
2) are you running?

This surprises me that you would have 3696736.



----- Original Message -----
From: "mats thunholm" <mats.thunholm@eu.apbiotech.com>
To: <jdelistml@jdelist.com>
Sent: Tuesday, February 27, 2001 6:01 AM
Subject: Re: Semaphore/Mutex Problem


>
> We installed that SAR (3696736) 2000-08-22.
>
> Are maybe my CPI values in the Server INI file to low ?
>
> [JDEIPC]
> maxNumberOfResources=1000
> startIPCKeyValue=2101
> avgResourceNameLength=15
> maxMsgqEntries=1024
> maxMsgqBytes=65536
> ipcTrace=0
>
> Thanks
> Mats Thunholm
>
>
--------------------------------------------------------------------------
-------------------------------
>
> Mats Thunholm
> APBiotech
> Uppsala, Sweden
> mats.thunholm@eu.apbiotech.com
>
--------------------------------------------------------------------------
-------------------------------

> > > *************************************************************
> > > This is the JDEList One World / XE Mailing List.
> > > Archives and information on how to SUBSCRIBE, and
> > > UNSUBSCRIBE can be found at http://www.JDELIST.com
> > > *************************************************************
> > >
> >
> >
> >
> >
> >
> > --------------------------
> > To view this thread, visit the JDEList forum at:
> >
>
http://198.144.193.139/cgi-bin/wwwthreads/showflat.pl?Cat=0&Board=OW&Number
> =
> 6235
> > *************************************************************
> > This is the JDEList One World / XE Mailing List.
> > Archives and information on how to SUBSCRIBE, and
> > UNSUBSCRIBE can be found at http://www.JDELIST.com
> > *************************************************************
> >
> >
> *************************************************************
> This is the JDEList One World / XE Mailing List.
> Archives and information on how to SUBSCRIBE, and
> UNSUBSCRIBE can be found at http://www.JDELIST.com
> *************************************************************
>
>
>
>
>
>
> B7332 SP10, AS400 V4R5
> Mats Thunholm
> APBiotech
> Uppsala, Sweden
>
> --------------------------
> To view this thread, visit the JDEList forum at:
>
http://198.144.193.139/cgi-bin/wwwthreads/showflat.pl?Cat=0&Board=OW&Number
=
6264
> *************************************************************
> This is the JDEList One World / XE Mailing List.
> Archives and information on how to SUBSCRIBE, and
> UNSUBSCRIBE can be found at http://www.JDELIST.com
> *************************************************************
>
>



JD Nowell
OW: B7332
ES: AS400 V4R4 CO: DB2/400 SP: 11.2
Users: 250 TSE Users: 100

--------------------------
To view this thread, visit the JDEList forum at:
http://198.144.193.139/cgi-bin/wwwthreads/showflat.pl?Cat=0&Board=OW&Number=6271

*************************************************************
This is the JDEList One World / XE Mailing List.
Archives and information on how to SUBSCRIBE, and
UNSUBSCRIBE can be found at http://www.JDELIST.com
*************************************************************






B7332 SP10, AS400 V4R5
Mats Thunholm
APBiotech
Uppsala, Sweden
 
Top