Re: alter table move partition - row cache lock

Looks like setting _kqr_optimistic_reads to TRUE eliminates the issue and the move of the partition completes very fast but not sure yet about the impact of changing it for 12.1 (even though in 19c is by default TRUE).

Thanks.

În sâm., 27 feb. 2021 la 05:12, Laurentiu Oprea <laurentiu.oprea06_at_gmail.com> a scris:

> Out of curiosity, has someone any idea what _kqr_optimistic_reads
> parameter does?
>
> I seen in 12.1 is false but in my 19c test database is true
>
> Thanks.
>
> În sâm., 27 feb. 2021 la 05:05, Laurentiu Oprea <
> laurentiu.oprea06_at_gmail.com> a scris:
>
>> Hello Mark,
>>
>> When connecting using a service, the parallel slaves will be spawned only
>> on the instance/instances where service is started. Parallel_force_local
>> will prevent parallel slaves from leaving the current instance.
>>
>> I haven't tried yet hanganalyze, I just started to look into it, I was
>> curious if someone faced a similar issue or had any clue.
>>
>> Thanks
>>
>> În vin., 26 feb. 2021 la 23:37, Lothar Flatz <l.flatz_at_bluewin.ch> a
>> scris:
>>
>>> Did you try a hanganalyze?
>>>
>>>
>>> Am 26.02.2021 um 13:59 schrieb Laurentiu Oprea:
>>>
>>> Hello,
>>>
>>> DB Version 12.0.1.2
>>>
>>> I noticed a strange situation and I`m curios if someone faced a similar
>>> situation or have any clue about the reason:
>>>
>>> While executing a partition move:
>>> ALTER TABLE t move PARTITION p TABLESPACE t ROW store COMPRESS advanced
>>> lob l store AS (COMPRESS medium TABLESPACE tl)
>>> UPDATE INDEXES PARALLEL 8;
>>>
>>> -> one query slave remain stuck into wait "row cache lock" (p2 = 0, p3
>>> =3)
>>> -> the blocker is actual query coordinator who is idle all this time
>>>
>>> Much appreciated,
>>> Laurentiu.
>>>
>>>
>>>

--
http://www.freelists.org/webpage/oracle-l