Thread: Kernel panic error

Kernel panic error

From
Robert.Farrugia@go.com.mt
Date:
Hi,

I have been using Postgres 7.0.3 for the last few weeks.  I also use LVM on
the 2.4 kernel.  I have very large tables (one of which is over 2GB).

My problem is that lately (the last week), when doing maintenance (normally
at night using a cronjob), the follwoing error has been repeatly given:

Incorrect number of segments after building list
nr_segments is 8
counted segments is 2
Flag 1 0
Segement 0xc5fa74a0, blocks 8, addr 0x4007ffff
Kernel panic: Thats all folks. Too dangerous to continue.

I have noticed yesterday that postgres gave the error below before the
kernel panic.  This error was given when vacuuming one of the smallest
tables in the database.
Error index_formtuple: data takes 21268 bytes: too big

Anyone has any ideas what is happening ?

Thanks
Robert


Re: Kernel panic error

From
Robert.Farrugia@go.com.mt
Date:
Hi,

I'm using the Logical Volume Manager with the 2.4.1 (sorry for forgetting
to specify the exact version).  What I am now suspecting is memory.  Once I
downgraded to 128MB (from 384MB), this night the error did not show up.

What showed up was the "Error index_formtuple: data takes 21268 bytes: too
big".  If anyone has any ideas on this, please share them.

Thanks for the interest
Robert





                    Denis Pugnere
                    <Denis.Pugnere@ig        To:     Robert.Farrugia@go.com.mt
                    h.cnrs.fr>               cc:
                                             Subject:     Re: Kernel panic error
                    16/02/2001 09:50






Yesterday, the 15 February 2001 at 10:12, Robert.Farrugia@go.com.mt wrote :

 | Hi,
 |
 | I have been using Postgres 7.0.3 for the last few weeks.  I also use LVM
on
 | the 2.4 kernel.  I have very large tables (one of which is over 2GB).

are you using RAID ?
the 2.4 kernel seems to have some panics in specific cases.
try 2.4.1

 |
 | My problem is that lately (the last week), when doing maintenance
(normally
 | at night using a cronjob), the follwoing error has been repeatly given:
 |
 | Incorrect number of segments after building list
 | nr_segments is 8
 | counted segments is 2
 | Flag 1 0
 | Segement 0xc5fa74a0, blocks 8, addr 0x4007ffff
 | Kernel panic: Thats all folks. Too dangerous to continue.
 |
 | I have noticed yesterday that postgres gave the error below before the
 | kernel panic.  This error was given when vacuuming one of the smallest
 | tables in the database.
 | Error index_formtuple: data takes 21268 bytes: too big
 |
 | Anyone has any ideas what is happening ?
 |
 | Thanks
 | Robert
 |
 |

Denis Pugnère
---
Denis.Pugnere@igh.cnrs.fr | IGH/CNRS UPR 1142, 141 Rue de la Cardonille
Tel : +33 (0)4 9961.9909  |     34396 Montpellier Cedex 5, France
Fax : +33 (0)4 9961.9901  |           http://www.igh.cnrs.fr






Re: Re: Kernel panic error

From
Peter Eisentraut
Date:
Robert.Farrugia@go.com.mt writes:

> What showed up was the "Error index_formtuple: data takes 21268 bytes: too
> big".  If anyone has any ideas on this, please share them.

It means your data is too big to fit into an index.

--
Peter Eisentraut      peter_e@gmx.net       http://yi.org/peter-e/


Re: Re: Kernel panic error

From
Tom Lane
Date:
Robert.Farrugia@go.com.mt writes:
> What showed up was the "Error index_formtuple: data takes 21268 bytes: too
> big".  If anyone has any ideas on this, please share them.

That says that you have a value too wide to fit in an index entry.  If
it was from data that fit before, then I think this must indicate that
data on-disk has gotten corrupted, causing some datum to appear longer
than it was --- and then when vacuum tries to rebuild the index entry
for that row, you get a failure.

In any case I'd say this is a consequence of your kernel-level problem.
It cannot be the cause.

            regards, tom lane