Saposcol dies after Hardware Information is required

[Former Member]

Hi,

We are facing a problem that hoppefully somebody could give us some ideas that could lead us to solve the problem.

As we are working with PI 7.1 the Saposcol service does not exists any more because now this is a process controled by a SAP service called SAPSERVICECTRL that is supposed to keep Saposcol up and running.

As part of the routine here, EWA's are generated every week and part of the process that Maintenance Package does every day is to connect PI from Solman to get info to latter create the EWA. Part of the Maintenance Package process is to get Hardware Information that is given by Saposcol which uses an OS process to create an file called hw_info.

The problem is as following:

We restart SAPSERVICECTRL and Saposcol starts and keeps up and recopilating OS info during all day long but when the Maintenance Package is executed at 4:00 am every day, Saposcol process dies and won´t restart until manually we restart SAPSERVICECTRL so while it is down no data is collected and in ST06 data is not complete and also EWA's for this system does not have information either.

We tried reinstalling SAPHOSTAGENT with the lattest patch with no luck, the same problem happened. If I open Saposcol -d and then execute de Saposcol command "ask Hardware" the same problem happens, Saposcol process dies after 20 seconds after Hardware Information is requested either from ST06 or from ask Hardware command.

This PI 7.1 server is running in MSCS Cluster in Windows 2003 x86 64bits.

One thing I notice was that when connecting with terminal service a Windows errors pops up giving an error like Saposcol Get Address Failed that leads me to think this is a Memory related problem. We have already from 3 weeks ago an OSS note and are still applying recomendations with no luck.

Have any of you experience a problem like this or have any idea about what could be happening here?

Your help will be appreciated.

[holger_stumm2 · Accepted solution]

Hi,

you need to open an OSS Call for this. My customer was facing the same issue and only a SAPNet Service call could solve it.

I cannot ellaborate on this, since it is customer/SAP internal until a fix is released .

Cheers

Holger

[Former Member · Accepted solution]

Hi,

Thanks for your imput, in fact we have an open High OSS case since November 18th for this problem but no luck until now so I would like to ask you, if this is possible for you, Do you remember what the title was for that OSS call your client wrote? or OSS internal message Number?. This is because we have Max Attention here and they can look for the message and give the SAP processor for our message the reference so they could probably fix this or give us a workaround.

At this point we had to delay until new notice a SAP Service that needs EWA reports in working order and this service is really important for the business. Some other SAP services will be affected also but we found the way to move on in those services using for some parts of it the QAS server or DEV server.

If you can help us with this we will apprecieate it a lot.

Regards,

[holger_stumm2 · Accepted solution]

please look at you email, I sended a private email. When the result is public, I will post it here for other people.

[Former Member]

As an update for this post, we got the last SAPHOSTAGENT that is supposed to handle the problem we have but first we had to apply the SAP Note 1375494 - SAP system does not start after applyingSAP kernel patch according to SAP suggestion.

Be aware that this note is mandatory for some SAP Kernels running on top of Windows 2003.

The SAPHOSTAGENT patch 59 includes the fixes fot the problem of Saposcol being closed after HW Info is required but for some reason I don´t understand yet, it did not work (SAPHOSTAGENT) after installation. In fact in this exact moment SAPHOSTAGENT is not available for PI. Probably this is due some update but I will wait until I get some more info to update this Post.

[Former Member]

Ok,

For all of you that could face this problem, the SAPHOSTAGENT patch 59 Did solved the problem after a Server reboot (Windows as you could imagine so the solution was that patched version.

Now some personal thougths about PI 7.1, this version is getting almost a new kernel version per week ! This system is working as productive in my client landscape for some time now but very restricted for obvious reasons.I think this is a version with lots of problems in Kernel so be aware of this after you install PI 7.1 and keep updating kernel as fast as you can.

Happy Holidays

[peter_dzurov]

Thanks for sharing such info. Good to know that.

[former_member66264]

No new entries are written to the log when it crashes?

It's strange that SAP can not find this problem.. do you have exotic hardware?

Sorry, I am a bit afraid its to much wintel related for me ( only work on unix)

I keep following the thread just in case ...

[Former Member]

You are right, there is no new log entries when Saposcol Crashes, Yesterday I found out that if I enter Saposcol -d (dialog) and then use ask HardwareXML command, then the file is created but Saposcol won't die. This should have something to do with OS (Windows) as per the way it dies.

SAP people told me that we should go up in SAPBasis component (3 more sp's) but I really doubt that this will do any good because I can´t see any relation between and OS related program and something that is running on top of a WorkProcess. The other problem is that at this season if we ask for an unplanified downtime with no warranty that this will solve anything is too risky for the executives to allow us to do it. This is a Huge Beer/Soda/food company so you can imagine the economic impact of taking down PI.

This is in HP Hardware not exotic at all

Anyway thanks for your interest and you can be sure that if I get the solution to this problem I will post it here right away.

[former_member66264]

HI,

Can you post thelogfiles dev_coll:

:

http://help.sap.com/saphelp_nw04/helpdata/en/c4/3a6c67505211d189550000e829fbbd/frameset.htm

Do you see anything in the os logs?

[Former Member]

Hi,

In OS Event Viewer there is no error altough when connecting using terminal server a screens shows up indicating that saposcol had a problem and tries to send the information to Microsoft. Looks like a memory address or so. About the dev coll content you asked this is it.

SAPOSCOL version COLL 21.02 710 - 21.06 NT 09/10/29, 64 bit, multithreaded, Non-Unicode

compiled at Nov 8 2009

systemid 562 (PC with Windows NT)

relno 7100

patch text COLL 21.02 710 - 21.06 NT 09/10/29

patchno 176

intno 20020600

running on (Windows server) Windows NT 5.2 3790 Service Pack 2 8x AMD64 Level 6 (Mod 23 Step 6)

chgno(local) 1097385 chgno(global) 1109817

PATCHES

DATE CHANGELIST PLATFORM PATCHTEXT

20081211 1030697 ALL Option -w support. Removed SizeOfRecord warning rt 130.

20090114 1036523 UNIX Log file permissions: 664.

20090203 1041527 UNIX Add. single instance check.

20090210 1042963 ALL Continue after EWA HW XML generation failure.

20090324 1052749 ALL Fix dynamic trace level switch.

20090414 1057723 ALL Usage of DIR_HOME as working dir.

20090512 1064065 UNIX Hardware XML permissions 660.

20090518 1065501 WINDOWS Support of VMware ESX 4.0 on Windows

20090617 1066953 WINDOWS Support of Hyper-V on Windows

20091009 1101740 UNIX Hardware XML permissions 664.

16:01:18 30.11.2009 LOG: "C:\Program Files\SAP\hostctrl\exe\saposcol.exe" -q -w60 pf="C:\Program Files\SAP\hostctrl\exe\host_profile"

16:01:18 30.11.2009 LOG: Profile : C:\Program Files\SAP\hostctrl\exe\host_profile

16:01:18 30.11.2009 LOG: Saposcol Version : [COLL 21.02 710 - 21.06 NT 09/10/29]

16:01:18 30.11.2009 LOG: Working directory : H:\usr\sap\PRFCLOG

16:01:18 30.11.2009 LOG: Allocate Counter Buffer [10000 Bytes]

16:01:18 30.11.2009 LOG: Allocate Instance Buffer [10000 Bytes]

16:01:18 30.11.2009 LOG: VIRT- Initialize virtual collection

16:01:18 30.11.2009 LOG: VIRT- VM: Hostname is XXXXXXX

16:01:18 30.11.2009 LOG: VIRT- Virtualization Software: HP, ProLiant BL460c G1.

16:01:18 30.11.2009 LOG: You can ignore :"Index of Title:[Disk Queue Length] not found" on Windows NT 4.0

16:01:18 30.11.2009 LOG: You can ignore :"Index of Title:[Disk Queue Length] not found" on Windows NT 4.0

16:01:18 30.11.2009 LOG: You can ignore :"Index of Title:[Disk Queue Length] not found" on Windows NT 4.0

16:01:18 30.11.2009 LOG: You can ignore :"Index of Title:[Disk Queue Length] not found" on Windows NT 4.0

16:01:18 30.11.2009 LOG: You can ignore :"Index of Title:[Disk Queue Length] not found" on Windows NT 4.0

16:01:18 30.11.2009 LOG: You can ignore :"Index of Title:[Disk Queue Length] not found" on Windows NT 4.0

16:01:18 30.11.2009 LOG: You can ignore :"Index of Title:[Disk Queue Length] not found" on Windows NT 4.0

16:01:18 30.11.2009 LOG: You can ignore :"Index of Title:[Disk Queue Length] not found" on Windows NT 4.0

16:01:18 30.11.2009 LOG: You can ignore :"Index of Title:[Disk Queue Length] not found" on Windows NT 4.0

16:01:18 30.11.2009 LOG: You can ignore :"Index of Title:[Disk Queue Length] not found" on Windows NT 4.0

16:01:18 30.11.2009 LOG: InitFsysDataWMI: allocating buffer for 50 NCFS objects

16:01:18 30.11.2009 LOG: InitFsysDataWMI: allocating name buffer of size 600

16:01:18 30.11.2009 LOG: using WMI to collect Filesystem data

16:01:18 30.11.2009 LOG: /* validation records */ 20

16:01:18 30.11.2009 LOG: /* disks * 25 */ + PX_disk_anz_max * 25 = 50 * 25 = 1250

16:01:18 30.11.2009 LOG: /* filesystems */ + PX_fsys_max_wmi * 25 = 50 * 25 = 1250

16:01:18 30.11.2009 LOG: /* cpu all */ + 25

16:01:18 30.11.2009 LOG: /* cpus */ + PX_scpu_anz = 8

16:01:18 30.11.2009 LOG: /* lans */ + PX_lan_anz_max * 25 = 50 * 25 = 1250

16:01:18 30.11.2009 LOG: /* for memory */ + 24

16:01:18 30.11.2009 LOG: /* top 40 processes */ + NBR_TOP = 40

16:01:18 30.11.2009 LOG: /* process list */ + MAX_PROCESS_CONF_ENTRIES = 50

16:01:18 30.11.2009 LOG: /* systemid */ + 1

16:01:18 30.11.2009 LOG: /* kernelparameter */ + kpar_anz * 2 = 0 * 2 = 0

16:01:18 30.11.2009 LOG: + spec_anz * 25 = 1 * 25 = 25

16:01:18 30.11.2009 LOG: + 25

16:01:18 30.11.2009 LOG: RecordNeeded: 3968

16:01:18 30.11.2009 LOG: Shared Memory Size: 427422.

16:01:18 30.11.2009 LOG: Connected to existing shared memory.

16:01:18 30.11.2009 LOG: Reused shared memory. Clearing contents.

16:01:21 30.11.2009 LOG: Collector daemon started

16:01:21 30.11.2009 LOG: read coll.put Thu Nov 26 11:59:52 2009

16:01:21 30.11.2009 LOG: Collector PID: 2036

16:02:21 30.11.2009 LOG: Process Monitoring active.

16:02:21 30.11.2009 LOG: searching for Process Monitoring Templates in H:\usr\sap\PRFCLOG\dev_proc

16:02:21 30.11.2009 LOG: searching for Process Monitoring Templates in H:\usr\sap\PRFCLOG\procmon\

16:02:21 30.11.2009 LOG: The following processes will be monitored:

Thanks ,

AC

[Former Member]

Hi AC

Can you please ensure you have both SAPHostExec & SAPHostControl services set to 'automatic' & configured to run using your SIDADM account.

For SAPHostExec, specify this in 'path to executable'

"C:\Program Files\SAP\hostctrl\exe\saphostexec.exe" pf="C:\Program Files\SAP\hostctrl\exe\saphostexec.exe"

For SAPHostControl, specify this in 'path to executable'

"C:\Program Files\SAP\hostctrl\exe\sapstartsrv.exe pf="C:\Program Files\SAP\hostctrl\exe\sapstartsrv.exe"

I hope this helps you.

Regards

Sekhar

[Former Member]

Thanks for your answer, but they are automatic and pointing to the correct path.

Both services keep started but for some reason Saposcol is getting down when you ask for HW Info if we don´t ask for it, then Saposcol will keep up forever.