Re: Blocker on checkpoint

Fred Toussi via Hsqldb-user <[email protected]> Thu, 14 Apr 2022 09:31:16 +0100
Newsgroups gmane.comp.java.hsqldb.user
Message-ID <[email protected]>
--===============4871906196163284051==
Content-Type: multipart/alternative;
 boundary=992aee31f2ff46a0b8e9fec761b6a94b

--992aee31f2ff46a0b8e9fec761b6a94b
Content-Type: text/plain;charset=utf-8
Content-Transfer-Encoding: quoted-printable

It is not clear what may be the cause. Enable SQL EVENT LOG LEVEL 2 and =
see if there is any indication.

In the HSLQDB Guide, it is mentioned that NIO use was reported be proble=
matic in some setups.

The latest versions of HSQLDB support table spaces. If you turn table sp=
aces on for a database with large cached tables, you probably do not nee=
d to do CHECKPOINT DEFRAG so often.

Fred

On Thu, Apr 14, 2022, at 07:48, S=C3=A9bastien P=C3=A9r=C3=A8s-Labourdet=
te wrote:
>=20
> Hi.
>=20
> Another instance is now causing same issue.
>=20
> Everything is fine for a certain amount of time as well as checking in=
stance individually in-process :
>  - checkpoint / defrag
>  - perform check all table index and fix
>  - shutdown
>=20
> At some point, unable to checkpoint / defrag properly.
> Disk becoming full after a month or so, unable to backup, hsqldb unabl=
e to sync io, databases becomes corrupted ....
>=20
> What may be causing defrag / checkpoint to hang like that ?
>=20
>=20
>=20
> *De : *Fred Toussi Via Hsqldb-user <[email protected]>
> *Envoy=C3=A9 : * Vendredi 8 Avril 2022 18:18:03
> *=C3=80 : *Hsqldb User List <[email protected]>
> *Cc : *Fred Toussi <[email protected]>
> *Objet : *Re: [Hsqldb-user] Blocker on checkpoint
>> There's obviusly an issue with the problematic instance.=20
>>=20
>> Execute SHUTDOWN on this particular instance and open it as an in-pro=
cess database using DatabaseManagerSwing then try "perform check all tab=
le index and fix". If it doesn't end, there is something wrong that it c=
annot fix. Also try SHUTDOWN SCRIPT. If it doesn't work either you can g=
et SupportWare and send the database for a possible fix.
>>=20
>> Fred
>>=20
>> On Fri, Apr 8, 2022, at 10:39, S=C3=A9bastien P=C3=A9r=C3=A8s-Labourd=
ette wrote:
>>=20
>>> Hi there.
>>>=20
>>> Use case is the following
>>>  - Hsqldb server 2.5.1 with 14 database instances, some 100% memory,
>>> some 100% cached.
>>>  - Background java job to perform checkpoint defrag on all instances.
>>>  - Job running well for 4/5 days.
>>>  - At some point, checkpoint defrag does never end on one specific i=
nstance
>>>  - Shutdown server does never end (waiting for the problematic insta=
nce
>>> to shutdown but never comes)
>>>  - Force to kill the server process
>>>  - Restart server and all goes well for the next 4/5 days
>>>=20
>>> The problematic instance does not mention any memory table (all cach=
ed)
>>>=20
>>> For this one, as well as some others, we have
>>>  - SET FILES WRITE DELAY FALSE
>>>=20
>>> As well as
>>>  - SET FILES BACKUP INCREMENT TRUE
>>>  - SET FILES NIO TRUE
>>>  - SET FILES NIO SIZE 1024
>>>=20
>>> Statements such as "perform check all table index and fix" does never
>>> end as well.
>>>=20
>>> Storage / hardware is the same for all 14 instances, database "size"=
 is
>>> 33Mb, other non-problematic databases goes beyond +600Mb
>>>=20
>>> Any clues ?
>>>=20
>>> --
>>> S=C3=A9bastien
>>>=20
>>>=20
>>> _______________________________________________
>>> Hsqldb-user mailing list
>>> [email protected]
>>> https://lists.sourceforge.net/lists/listinfo/hsqldb-user
>>>=20
>>> Attachments:
>>> * signature.asc
>>>=20
>> _______________________________________________
>> Hsqldb-user mailing list
>> [email protected]
>> https://lists.sourceforge.net/lists/listinfo/hsqldb-user
>>=20
>=20
> _______________________________________________
> Hsqldb-user mailing list
> [email protected]
> https://lists.sourceforge.net/lists/listinfo/hsqldb-user
>=20
>=20
> *Attachments:*
>  * signature.asc

--992aee31f2ff46a0b8e9fec761b6a94b
Content-Type: text/html;charset=utf-8
Content-Transfer-Encoding: quoted-printable

<!DOCTYPE html><html><head><title></title><style type=3D"text/css">p.Mso=
Normal,p.MsoNoSpacing{margin:0}</style></head><body><div style=3D"font-f=
amily:Arial;">It is not clear what may be the cause. Enable SQL EVENT LO=
G LEVEL 2 and see if there is any indication.<br></div><div style=3D"fon=
t-family:Arial;"><br></div><div style=3D"font-family:Arial;">In the HSLQ=
DB Guide, it is mentioned that NIO use was reported be problematic in so=
me setups.<br></div><div style=3D"font-family:Arial;"><br></div><div sty=
le=3D"font-family:Arial;">The latest versions of HSQLDB support table sp=
aces. If you turn table spaces on for a database with large cached table=
s, you probably do not need to do CHECKPOINT DEFRAG so often.<br></div><=
div style=3D"font-family:Arial;"><br></div><div style=3D"font-family:Ari=
al;">Fred<br></div><div style=3D"font-family:Arial;"><br></div><div>On T=
hu, Apr 14, 2022, at 07:48, S=C3=A9bastien P=C3=A9r=C3=A8s-Labourdette w=
rote:<br></div><blockquote type=3D"cite" id=3D"qt" style=3D""><div id=3D=
"qt-smartTemplate4-template"><br></div><div>Hi.<br></div><div> <br></div=
><div> Another instance is now causing same issue.<br></div><div> <br></=
div><div> Everything is fine for a certain amount of time as well as che=
cking
    instance individually in-process :<br></div><div> &nbsp;- checkpoint=
 / defrag<br></div><div> &nbsp;- perform check all table index and fix<b=
r></div><div> &nbsp;- shutdown<br></div><div> <br></div><div> At some po=
int, unable to checkpoint / defrag properly.<br></div><div> Disk becomin=
g full after a month or so, unable to backup, hsqldb
    unable to sync io, databases becomes corrupted ....<br></div><div> <=
br></div><div> What may be causing defrag / checkpoint to hang like that=
 ?<br></div><div> <br></div><div> <br></div><div> <br></div><div id=3D"q=
t-smartTemplate4-quoteHeader"><div style=3D"font-size:9pt;font-family:Ar=
ial, Helvetica;padding-bottom:1px;color:rgb(0, 0, 0);padding-top:1px;"><=
div><hr size=3D"1"><b>De : </b>Fred Toussi Via Hsqldb-user <a class=3D"q=
t-moz-txt-link-rfc2396E" href=3D"mailto:[email protected]=
t">&lt;[email protected]&gt;</a><br></div><div> <b>Envoy=
=C3=A9 : </b> Vendredi 8 Avril 2022 18:18:03<br></div><div> <b>=C3=80 : =
</b>Hsqldb User List <a class=3D"qt-moz-txt-link-rfc2396E" href=3D"mailt=
o:[email protected]">&lt;[email protected]=
et&gt;</a><br></div><div> <b>Cc : </b>Fred Toussi <a class=3D"qt-moz-txt=
-link-rfc2396E" href=3D"mailto:[email protected]">&lt;fredt@us=
ers.sourceforge.net&gt;</a><br></div><div> <b>Objet : </b>Re: [Hsqldb-us=
er] Blocker on checkpoint<br></div></div></div><blockquote type=3D"cite"=
 cite=3D"mid:[email protected]"><pre=
 class=3D"qt-moz-quote-pre">There's obviusly an issue with the problemat=
ic instance.=20

Execute SHUTDOWN on this particular instance and open it as an in-proces=
s database using DatabaseManagerSwing then try "perform check all table =
index and fix". If it doesn't end, there is something wrong that it cann=
ot fix. Also try SHUTDOWN SCRIPT. If it doesn't work either you can get =
SupportWare and send the database for a possible fix.

Fred

On Fri, Apr 8, 2022, at 10:39, S=C3=A9bastien P=C3=A9r=C3=A8s-Labourdett=
e wrote:
<br></pre><blockquote type=3D"cite"><pre class=3D"qt-moz-quote-pre">Hi t=
here.

Use case is the following
&nbsp;- Hsqldb server 2.5.1 with 14 database instances, some 100% memory,
some 100% cached.
&nbsp;- Background java job to perform checkpoint defrag on all instance=
s.
&nbsp;- Job running well for 4/5 days.
&nbsp;- At some point, checkpoint defrag does never end on one specific =
instance
&nbsp;- Shutdown server does never end (waiting for the problematic inst=
ance
to shutdown but never comes)
&nbsp;- Force to kill the server process
&nbsp;- Restart server and all goes well for the next 4/5 days

The problematic instance does not mention any memory table (all cached)

For this one, as well as some others, we have
&nbsp;- SET FILES WRITE DELAY FALSE

As well as
&nbsp;- SET FILES BACKUP INCREMENT TRUE
&nbsp;- SET FILES NIO TRUE
&nbsp;- SET FILES NIO SIZE 1024

Statements such as "perform check all table index and fix" does never
end as well.

Storage / hardware is the same for all 14 instances, database "size" is
33Mb, other non-problematic databases goes beyond +600Mb

Any clues ?

--
S=C3=A9bastien


_______________________________________________
Hsqldb-user mailing list
<a class=3D"qt-moz-txt-link-abbreviated" href=3D"mailto:Hsqldb-user@list=
s.sourceforge.net">[email protected]</a>
<a class=3D"qt-moz-txt-link-freetext" href=3D"https://lists.sourceforge.=
net/lists/listinfo/hsqldb-user">https://lists.sourceforge.net/lists/list=
info/hsqldb-user</a>

Attachments:
* signature.asc
<br></pre></blockquote><pre class=3D"qt-moz-quote-pre">_________________=
______________________________
Hsqldb-user mailing list
<a class=3D"qt-moz-txt-link-abbreviated" href=3D"mailto:Hsqldb-user@list=
s.sourceforge.net">[email protected]</a>
<a class=3D"qt-moz-txt-link-freetext" href=3D"https://lists.sourceforge.=
net/lists/listinfo/hsqldb-user">https://lists.sourceforge.net/lists/list=
info/hsqldb-user</a>
<br></pre></blockquote><div><br></div><div>_____________________________=
__________________<br></div><div>Hsqldb-user mailing list<br></div><div>=
<a href=3D"mailto:[email protected]">[email protected]=
ourceforge.net</a><br></div><div><a href=3D"https://lists.sourceforge.ne=
t/lists/listinfo/hsqldb-user">https://lists.sourceforge.net/lists/listin=
fo/hsqldb-user</a><br></div><div><br></div><div><br></div><div><b>Attach=
ments:</b><br></div><ul><li>signature.asc<br></li></ul></blockquote><div=
 style=3D"font-family:Arial;"><br></div></body></html>
--992aee31f2ff46a0b8e9fec761b6a94b--


--===============4871906196163284051==
Content-Type: text/plain; charset="us-ascii"
MIME-Version: 1.0
Content-Transfer-Encoding: 7bit
Content-Disposition: inline


--===============4871906196163284051==
Content-Type: text/plain; charset="us-ascii"
MIME-Version: 1.0
Content-Transfer-Encoding: 7bit
Content-Disposition: inline

_______________________________________________
Hsqldb-user mailing list
[email protected]
https://lists.sourceforge.net/lists/listinfo/hsqldb-user

--===============4871906196163284051==--