PEM: Error in REINDEX DATABASE PEM

[Polem]

Hi,

After fixing corrupted data in pem.alert and pem.alert_status(see previous post)

pem=# reindex (verbose) table  pem.alert;

ERROR:  failed to find parent tuple for heap-only tuple at (290,6) in table "alert"

pem=# delete from pem.alert where ctid = '(290,6)';

DELETE 1

pem=# reindex (verbose) table  pem.alert;

INFO:  index "alert_pkey" was reindexed

DETAIL:  CPU: user: 0.00 s, system: 0.00 s, elapsed: 0.00 s

INFO:  index "alert_name_object_type_uniq" was reindexed

DETAIL:  CPU: user: 0.00 s, system: 0.00 s, elapsed: 0.00 s

INFO:  index "pg_toast_19174_index" was reindexed

DETAIL:  CPU: user: 0.00 s, system: 0.00 s, elapsed: 0.00 s

Now I get another error in:

pem=# reindex (verbose) database pem;

I get this error:

ERROR:  could not create unique index "probe_schedule_pkey"
DETAIL:  Key (probe_id, parameter_value_list)=(19, {12,postgres}) is duplicated.

Kindly,

Polem

PS:

Turnoff 5432 port to block all agent during the REINDEX command to validate all data

[Polem]

Hi,

I was able to delete the duplicate records but now I found errors in pemdata.index_statistics.

Can I truncate this table?

Kindly

Polem

[dhananjayjejur]

Hi Paul,

As you are facing corruption in your PEM database which is mentioned via PostgresRocks another thread.

i.e. : https://postgresrocks.enterprisedb.com/t5/Postgres-Tools/Error-in-PEM-worker-on-PEM-server/m-p/2320#...

Where we have requested you re-setup the PEM instead of truncating alerts table. As it is not recommended.

Hope this answers your question.