transtate = 0 ( implies my transaction is in progress ) when it struck with signal 11 stack trace
↧
Re: Signal 11 infected
↧
Re: The log segment in the target database 'BIS' is too small (56320 MB) to accommodate the log segment from the dumped database (102400 MB).
Hi All,
Increased Log space more than 102400 MB and the problem has been resolved.
1> disk resize name='BIS_log_001', size= "55G"
2> go
1> alter database BIS log on BIS_log_001 = "55G"
2> go
Regards,
Karthik
↧
↧
Re: Signal 11 infected
I can give some basic guidelines. Generally the thing to do is figure out from the stacktrace what the most relevant function names are, and then use them as keywords to search the knowledge base.
The first place to look is at the very top, which usually gives the exact address the fault occurred on. In this case, that function name is "kbfusage". The ".fdpr.chunk.2" is not actually part of the function name, it is an artifact added by the feedback-optimized (FBO) optimizer. [Learning what to discard is part of the steep learning curve for learning to read stacktraces].
Current process (0x1525017b) infected with signal 11 (SIGSEGV)
Address 0x00000001007a0808 (kbfusage.fdpr.chunk.2+0xbc), siginfo (code, address) = (50, 0x0300003161000220)
Then we can go to the main stack. The very bottom usually has the connection handler invoked when the client first logs in (although there is a limit to the number of functions on the stack ASE will output, so if the stack is very deep the output may start in the middle rather than the beginning)
tds_recv_language is the server receiving the SQL text from the client.
sequencer indicates that the sql has been parsed and compiled into a list of compiled statements that are being executed in sequence.
Other commonly seen function names here are normalize and s_compile.
However, these functions towards the bottom are usually not important - but reading from the bottom up can give you a sense of what ASE is currently doing. Function names useful for keywords occur close to the top, but not the very top. Everything from kisignal() up has to do with handling the signal and printing out the stacktrace. In this case, the stack output show kisignal being called from kbfusage, which is the same function we saw on the initial address line. However, this doesn't always match, which is why I check the address line first. The first ~3 functions under kisignal are usually the best choices for the keyword. Sometimes the first or second are actually too generic (for example, lots of processes might call alloc(), but the bug is that the function that calls alloc() is passing in a bad pointer, so to be a good search you need to include the caller.). It is also sometimes the case that the first few functions are too specific, and some flaw in their caller might manifest in different ways in other subfunctions - so if you don't find a hit on the highest functions and they look pretty specific, it can pay to search for just the caller.)
pc: 0x0000000100026acc pcstkwalk+0x654()
pc: 0x0000000100025c40 ucstkgentrace+0x178()
pc: 0x0000000100025070 ucbacktrace+0xa8()
pc: 0x000000010011c084 terminate_process.fdpr.chunk.4+0x240()
pc: 0x0000000101bed0a0 kisignal+0x2e8()
pc: 0x00000001007a0808 kbfusage.fdpr.chunk.2+0xbc()
pc: 0x00000001005c42bc ubfprop.fdpr.chunk.3+0x8()
pc: 0x000000010007c87c kbprop+0xac()
pc: 0x0000000102a4ce6c mda__MemUsg_one_FragmentPool+0xc4()
pc: 0x0000000102a4c25c mda__MemUsg_FragmentPool+0x12c()
pc: 0x0000000102a4bb34 mda_populate_monMemoryUsage+0x1dc()
Anyway, for me searching for "kbfusage mda_populate_monMemoryUsage" found the bug, but this was using an internal search tool. Unfortunately there isn't yet a published KBA for this bug, so (prior to this SCN article) you probably couldn't have found it even using the same keywords. It is worth a try, though.
Cheers,
-bret
↧
Re: Signal 11 infected
True,I totally agree. This is already in place with SAP Support Colleagues.
My intention was to understand , or rather to discuss on many other ways of analyzing stack traces. Sharing your experiences and knowledge on how you do a basic interpretation of stack traces, which many of the colleagues in SAP ASE world still finds it very challenging.
Kind Regards,
Ajith Prabhakaran
↧
DB Migration from oracle to SAP ASE
Dear All,
We are in the planning of migrating our SAP systems from Oracle DB to SAP ASE DB. this can be done using our normal system copy procedure of Import and export option.
But we are trying to use the option of DMO of SUM for SAP ASE migration without creating new target machine (i.e. install the ASE db in the source system itself and complete the migration and later remove the oracle from it). please clarify me whether this scenario is possible.
if it possible, we will avoid the new system installation.
Thanks and Regards
Balaji
↧
↧
Re: Signal 11 infected
Many Thanks Bret. That was very informative and in depth explanatory. Truly it was very helpful, on how each breakups meant, and digging further more. I am sure, this would be very beneficial for readers and followers of SCN Articles.
I was almost close in the search. I ended up with
Note 1981791:SAP ASE hit infected with 11 in kbfusage
I maybe didn't look closer with the keyword "mda_populate_monMemoryUsage"
I got mislead with all the "Le" terms populated in the logs.
0:0017:00000:00433:2015/12/18 06:30:00.78 kernel pc: 0x00000001023889ec LeScanOp::_LeOpNext.fdpr.chunk.7(ExeCtxt&)+0x1a8()
00:0017:00000:00433:2015/12/18 06:30:00.79 kernel pc: 0x000000010063f8a0 LeInsertOp::_LeOpNext(ExeCtxt&)+0x1b8()
00:0017:00000:00433:2015/12/18 06:30:00.79 kernel pc: 0x0000000100629f6c LeEmitNoSndOp::_LeOpNext(ExeCtxt&)+0x2bc()
00:0017:00000:00433:2015/12/18 06:30:00.79 kernel pc: 0x00000001006296c0 LePlanNext+0x178()
↧
Re: Signal 11 infected
Thanks Bret !!!!
I think you are spot on.
↧
Re: Update Statistics - SAP ASE
1> select ObjectName=convert(varchar(20),object_name(p.id))
2> ,PartName=convert(varchar(20),p.name)
3> ,ColName=convert(varchar(20),c.name)
4> ,HistDt=convert(varchar(10),s.moddate,101)
5> ,Smp_Pct=str(s.c13,8)
6> ,Datachg=datachange("SAPSR3."+object_name(p.id),p.name,c.name)
7> from syspartitions p INNER JOIN syscolumns c
8> ON p.id = c.id
9> and p.indid between 0 and 1
10> LEFT JOIN sysstatistics s
11> ON c.id = s.id
12> and p.id = s.id
13> and s.formatid = 100
14> and s.statid = 0
15> and s.sequence = 1
16> and convert(varbinary,c.colid) = s.colidarray
17> and datalength(s.colidarray) <= 2
18> where object_name(p.id) = '/BI0/F0SMD_PE2H'
19> order by object_name(p.id),p.name,c.colid
20> go
Msg 207, Level 16, State 4:
Server 'S30', Line 1:
Invalid column name 'SAPSR3.'.
Hi Keven, just pasting the result. Please tell me, what needs to be corrected here.
↧
Re: DB Migration from oracle to SAP ASE
Hi Balaji
SUM is used to upgrade a system (& DMO option to upgrade + migrate to Hana).
SUM could not be used to perform a heterogeneous system copy (to migrate from Oracle to ASE), this should be done using SWPM.
Best regards
↧
↧
Re: DB Migration from oracle to SAP ASE
Hi Kervadec,
Thanks for the clarification, one more thing, is the same method we used to upgrade a system (& DMO option to upgrade + migrate to Hana) can be used to sybase migration i.e. target db as sybase? or DMO using SUM available only for HANA?
Regards
Balaji
↧
Re: ERROR: CLI error trying to establish connection: [unixODBC][Driver Manager] Can't open lib
Hi
The error says: file no found
Have you tried to look for this library file ? Does it exist?
Mabye it exists in other localization or has different name...
Can you also post the entries of /etc/odbcinst.ini ? (I assume you are using linux enviroment)
Regards.
--
Marcin
↧
Re: DB Migration from oracle to SAP ASE
Sorry but to my knowledge DMO is only for Hana...
Check Boris Rubarth's answer in the wiki.
indeed this procedure is exclusively for SAP HANA DB as target db only.
Database Migration Option (DMO) of SUM - Introduction
You can still use SWPM to perform a copy + upgrade on the same system :
1) Export DB using SWPM, uninstall Oracle & SAP
2) Install ASE and create a new SAP instance with SWPM using the export.
3) upgrade
You could also change the order and perform the upgrade before the copy, but no ways to perform upgrade & DB change in one pass with ASE.
Regards
↧
Re: Update Statistics - SAP ASE
include all statements that I posted please. I don't see the first part:
set quoted_identifier on
Before the SQL statement.
↧
↧
SAP ERP on ASE to Cluster
Hello
One of my clients wants to switch to cluster environment.
Their DB server reside on Windows 2012 R2 and SAP ASE 15.7 SP135. They have six additional application servers. DB server has 64 CPU cores and 256 GB RAM with 2.5 TB of DB space occupied. Application servers have 16 CPU and 64 GB RAM.
How can I switch it to MSCS environment with minimal downtime? What is the best practice about it?
I know I can do with system copy but it will take too much time (around 4 days).
Best regards
↧
Re: Signal 11 infected
ASE was originally written in the c language. The optimizer (query compiler) and execution engine were subsequently rewritten in c++. When you see function names with the double colons ( '::'), you are in the c++ functions (technically class methods). In general I think it is harder to figure out just what ASE is doing from these names - and unless the failure is actually occurring in one of them there usually isn't much need to figure it out. These names are also more likely to be munged up by the FBO optimizer, and not just by appending ".fdpr.chunk" at the end. In general, if you are searching for these start by searching for the whole "class::method" string but not including the offset value (the '+0x<hexvalue>()') number, which indicates how many bytes we are into the assembly language for the function or method.
↧
Re: signal 11 (SIGSEGV)
Hi Bret,
How to know which replication we are using in sybase replication server (like. warm standby or table level ) is there any specific commands. please provide the commands ans process.
↧
Re: Update Statistics - SAP ASE
Hi Kevin,
set quoted_identifier on, was the culprit. When i removed it , I got the output .
1> select ObjectName=convert(varchar(20),object_name(p.id))
2> ,PartName=convert(varchar(20),p.name)
3> ,ColName=convert(varchar(20),c.name)
4> ,HistDt=convert(varchar(10),s.moddate,101)
5> ,Smp_Pct=str(s.c13,8)
6> ,Datachg=datachange("SAPSR3."+object_name(p.id),p.name,c.name)
7> from syspartitions p INNER JOIN syscolumns c
8> ON p.id = c.id
9> and p.indid between 0 and 1
10> LEFT JOIN sysstatistics s
11> ON c.id = s.id
12> and p.id = s.id
13> and s.formatid = 100
14> and s.statid = 0
15> and s.sequence = 1
16> and convert(varbinary,c.colid) = s.colidarray
17> and datalength(s.colidarray) <= 2
18> where object_name(p.id) = '/BI0/F0SMD_PE2H'
19> order by object_name(p.id),p.name,c.colid
20> go | tail -4
/BI0/F0SMD_PE2H /BI0/F0SMD_PE2H_0000 SMD_KEY 12/17/2015 NULL 0.000000
/BI0/F0SMD_PE2H /BI0/F0SMD_PE2H_0000 SMD_KEY 12/17/2015 NULL 0.000000
(4812285 rows affected)
The output is very very huge.
I took the top 10 results:
which is
ObjectName PartName ColName HistDt Smp_Pct Datachg
-------------------- -------------------- -------------------- ---------- -------- ---------------------------
/BI0/F0SMD_PE2H /BI0/F0SMD_PE2H_0000 KEY_0SMD_PE2HP 02/06/2014 NULL 0.000000
/BI0/F0SMD_PE2H /BI0/F0SMD_PE2H_0000 KEY_0SMD_PE2HP 02/06/2014 NULL 0.000000
/BI0/F0SMD_PE2H /BI0/F0SMD_PE2H_0000 KEY_0SMD_PE2HP 02/06/2014 NULL 0.000000
/BI0/F0SMD_PE2H /BI0/F0SMD_PE2H_0000 KEY_0SMD_PE2HP 02/06/2014 NULL 0.000000
/BI0/F0SMD_PE2H /BI0/F0SMD_PE2H_0000 KEY_0SMD_PE2HP 02/06/2014 NULL 0.000000
/BI0/F0SMD_PE2H /BI0/F0SMD_PE2H_0000 KEY_0SMD_PE2HP 02/06/2014 NULL 0.000000
/BI0/F0SMD_PE2H /BI0/F0SMD_PE2H_0000 KEY_0SMD_PE2HP 03/24/2014 NULL 0.000000
/BI0/F0SMD_PE2H /BI0/F0SMD_PE2H_0000 KEY_0SMD_PE2HP 03/24/2014 NULL 0.000000
/BI0/F0SMD_PE2H /BI0/F0SMD_PE2H_0000 KEY_0SMD_PE2HP 03/24/2014 NULL 0.000000
/BI0/F0SMD_PE2H /BI0/F0SMD_PE2H_0000 KEY_0SMD_PE2HP 03/24/2014 NULL 0.000000
↧
↧
Re: Signal 11 infected
Thanks a lots Bret.
That was huge a lot of very useful and valuable information.
Sincerely appreciate your efforts to fetch these info and much importantly sharing them, giving proper explanations etc: -
Thanks again Bret.
Coming to the stact traces..
How about the handler messages towards the end of the stack trace.
Handler pc: 0x00000001007904e4 le_execerr installed by the following function:-]
00:0017:00000:00433:2015/12/18 06:30:00.80 kernel pc: 0x00000001006277b0 exec_lava+0x2d0()
00:0017:00000:00433:2015/12/18 06:30:00.80 kernel pc: 0x0000000100630bd4 s_execute.fdpr.chunk.32+0x2e4()
00:0017:00000:00433:2015/12/18 06:30:00.80 kernel [Handler pc: 0x00000001007903c4 hdl_stack installed by the following function:-]
00:0017:00000:00433:2015/12/18 06:30:00.81 kernel [Handler pc: 0x0000000100899ce8 s_handle installed by the following function:-]
00:0017:00000:00433:2015/12/18 06:30:00.82 kernel pc: 0x000000010061bba8 sequencer+0x480()
00:0017:00000:00433:2015/12/18 06:30:00.83 kernel [Handler pc: 0x00000001015be514 intsql__handle installed by the following function:-]
00:0017:00000:00433:2015/12/18 06:30:00.83 kernel pc: 0x000000010071be08 internal_sql+0x4b4()
00:0017:00000:00433:2015/12/18 06:30:00.83 kernel pc: 0x000000010071b71c s_execute.fdpr.chunk.55+0x54()
00:0017:00000:00433:2015/12/18 06:30:00.83 kernel [Handler pc: 0x00000001007903c4 hdl_stack installed by the following function:-]
00:0017:00000:00433:2015/12/18 06:30:00.83 kernel [Handler pc: 0x0000000100899ce8 s_handle installed by the following function:-]
00:0017:00000:00433:2015/12/18 06:30:00.83 kernel pc: 0x000000010061bba8 sequencer+0x480()
00:0017:00000:00433:2015/12/18 06:30:00.83 kernel pc: 0x0000000100612acc execproc+0x450()
00:0017:00000:00433:2015/12/18 06:30:00.83 kernel pc: 0x00000001006e2864 s_execute.fdpr.chunk.43+0xcc()
00:0017:00000:00433:2015/12/18 06:30:00.83 kernel [Handler pc: 0x00000001007903c4 hdl_stack installed by the following function:-]
00:0017:00000:00433:2015/12/18 06:30:00.83 kernel [Handler pc: 0x0000000100899ce8 s_handle installed by the following function:-]
00:0017:00000:00433:2015/12/18 06:30:00.83 kernel pc: 0x000000010061bba8 sequencer+0x480()
00:0017:00000:00433:2015/12/18 06:30:00.84 kernel pc: 0x000000010071d3c4 tdsrecv_language+0x258()
00:0017:00000:00433:2015/12/18 06:30:00.84 kernel [Handler pc: 0x00000001007747a4 ut_handle installed by the following function:-]
00:0017:00000:00433:2015/12/18 06:30:00.84 kernel pc: 0x000000010071cfc0 conn_hdlr.fdpr.chunk.12+0xc4()
00:0017:00000:00433:2015/12/18 06:30:00.84 kernel end of stack trace, spid 433, kpid 354746747, suid 4
What do they intend?
Kind Regards,
Ajith Prabhakaran
↧
Re: Update Statistics - SAP ASE
Dear Colleagues,
As said. This table is having 650 partitions ,. The index size being 250 gigs and data size 10 gigs.
Since the update statistics do not look like doing any much better on the statistics. Would you recommend to delete statistics, and then run the update all statistics on this table.
My concerns:
How bad would this affect the business? ( During the process of delete statistics and re run of update statistics )
Which Update Statistics command would the best suit, to run here , in that case. ( After delete statistics )
My schedule ( dbacockpit) jobs often, give the following error stating unable to write statistics for this table.
"0:0010:00000:00011:2015/12/28 03:45:31.76 server Unable to write statistics for object 1043999194 in database 4. Please run update statistics on this table after loading this database or prior to accessing the table in the loaded database."
Please Advise.
Kind Regards,
Ajith Prabhakaran
↧
Re: Unable to connect using default characterset. Attempting connection with iso_1
Hi PK,
This message is appearing due to the JINI configuration. The issue has been raised under CR # 788896 and fixed in ASE 15.7 SP 136.
I suggest you to check with applying the latest patch and see if this is helpful.
~Gaurav
↧