| Age | Commit message (Collapse) | Author |
|---|
|
The old wording could be interpreted as allowing several actions to be
executed in parallel.
Close #115.
|
|
checks
Close #81.
|
|
UNKNOWN is not so much useful in the context of basic connection checks.
(The result remains UNKNOWN in case version() returns something fishy.)
Close #100.
|
|
Close #63.
|
|
In passing, make check_postgres.pl executable
Close #74.
|
|
|
|
|
|
|
|
... and add a changelog entry for the same_schema fixes by Glyn.
|
|
As per the manual:
"To exclude objects of a certain type by a regular expression against their name, use "noname=regex"."
However the actual check in the script would only filter on an exact match; switched to regex.
|
|
owned by group roles in commit d55287f1064c615ce2b0eb3cc3e2f6261f8070c4
|
|
last_value check.
Fixes for changes made to same_schema in commit f8145cd902b2c1e01bc211bbfc04737c5fee79f9
* Amend constraint unit tests in t/02_same_schema.t to take into account we now return the table name
* Reinstate checking of a sequences last_val, and instead allow users of asynchronous replication to override with '--assume-async' option
|
|
operators.
|
|
constraint triggers
These triggers have a name containting the triggers oid which mismatches on logical replicas; so to exclude these just ensure tgconstrindid is zero.
|
|
|
|
This relates to a change in the character representing a simple match
on a foreign key stored in the confmatchtype field changing from 'u'
to 's' between postgresql 9.2 and 9.3. Ref:
http://www.postgresql.org/docs/9.2/static/catalog-pg-constraint.html
http://www.postgresql.org/docs/9.3/static/catalog-pg-constraint.html
|
|
|
|
rather than pg_user.
This resolves issues where objects owned by a group role are missing in various checks (check_same_schema, check_commitratio, check_hitratio), or detail is missing (check_database_size), or results not filtered as intended (check_last_vacuum_analyze, check_relation_size).
Conflicts:
check_postgres.pl
|
|
operators - Use text descriptions for result type and operands; ignore related oids
triggers - Include table name in key to prevent false positive when another table has a trigger with the same name. Use textual trigger definition; ignore tgqual represe
functions - Use text function definition; ignore oid based types and defaults.
constraints - Include table name in key to prevent false positive when another table has a constraint with the same name. Use text constraint definition; ignore oid based values.
sequence - Ignore last_value; I'm confused as to why we'd want to know this on a schema check. But if required could be added back in with an optional filter instead.
indexes - Ignore reltablespace oid; we check the tablespace name anyway. Ignore indkey attnum which can differ when tables have been altered differently but are otherwis
|
|
|
|
"Delay" is measured as size of transaction logs retained for the slot e.g:
check_postgres_replication_slots -db=TEST -H=192.168.0.106 -warning=32M -critical=64M
|
|
When check_replicate_row tested for valid values before starting the check, if the actual value returned was 0 it would evaluate to false, resulting in the error:
"Cannot test replication: values are not the same"
This is resolved by simply checking if the value is defined before assigning it's value a blank string instead.
|
|
|
|
http://stackoverflow.com/questions/550258/does-the-o-modifier-for-perl-regular-expressions-still-provide-any-benefit
|
|
pg_indexes_size doesn't exist either on 8.4, but that action is new, so
let's not care for now.
Also fix t/05_docs.t for the newly added actions.
|
|
Close: #75
|
|
Change table_size to use pg_table_size(), i.e. to include the TOAST
table size in the numbers reported. Add new actions indexes_size and
total_relation_size, using the respective pg_indexes_size() and
pg_total_relation_size() functions. All size checks will now also check
materialized views where applicable.
Close: #32, #83
|
|
Close #80
|
|
The extra space result in two spaces between the metrics and its incorrectly parse by centreon.
|
|
|
|
If a query contains the disabled word (.e.g. select 1 as disabled) "check_postgres.pl --action txn_idle" exits with
```UNKNOWN: No queries - is stats_command_string or track_activities off?```
If you set track_activities to off, the status field contains 'disabled' (at least on 9.3)
|
|
|
|
Starting in PostgreSQL 9.6, psql -c no longer implies -X, so it has to
be added explicitly in order to avoid customizations in .psqlrc
interfering with the output parsing.
|
|
The main PostgreSQL website is now https only. The versions.rss check
still works by following the redirect, but it's more efficient to hit
the https URL directly.
In passing, also update all the direct links in the documentation to be
https.
|
|
same-schema action will use to store its files.
|
|
|
|
|
|
The wrapper function around pg_ls_dir() must hardcode the pg_xlog or
pgxlog/archive_status path, and should take no arguments. It must be
created as a superuser with SECURITY DEFINER.
|
|
|
|
... and note author of last commit in changelog
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
