When running "<PHANTOM_HOME>/bin/phenv ibackup --setup" in Splunk SOAR (On-Premises), users may encounter the following error related to WAL archiving:
Command '['pgbackrest', '--stanza=phantom', '--config=/opt/phantom/etc/pgbackrest.conf', '--log-level-console=info', '--log-level-file=info', 'check']' returned non-zero exit status 82.
Output: 2023-07-25 21:07:43.772 P00 INFO: check command begin 2.28: --config=/opt/phantom/etc/pgbackrest.conf --log-level-console=info --log-level-file=info --log-path=/opt/phantom/var/log/phantom/backup --pg1-path=/opt/phantom/data/db --pg1-socket-path=/tmp --repo1-path=/opt/phantom/data/ibackup/repo/pg --stanza=phantom
ERROR: [082]: WAL segment 00000004000001620000003D was not archived before the 60000ms timeout
HINT: check the archive_command to ensure that all options are correct (especially --stanza).
HINT: check the PostgreSQL server log for errors.
2023-07-25 21:08:44.703 P00 INFO: check command end: aborted with exception [082]
This issue indicates that pgBackRest timed out while waiting for PostgreSQL to archive a WAL segment. It may occur due to slow disk I/O, delayed archive operations, a temporary system bottleneck or missing backup file.
This article provides two potential solutions:
Increasing the pgBackRest timeout (recommended first step)
By default, pgBackRest expects WAL segments to be archived within 60 seconds (archive-timeout = 60). If the archive operation is slow or temporarily blocked, the ibackup setup may fail with the above timeout error. In most SOAR deployments, this timeout is sufficient. However, if the system is under high load or the disk I/O is slow, the archive process may take longer than 60 seconds.
There could be accumulated large amounts of Postgresql WAL files under <PHANTOM_HOME>/data/db/pg_wal folders. And some of WAL files were not being archived correctly. You can verify the issue by reviewing:
The pgbackrest logs under <PHANTOM_HOME>/var/log/phantom/backup/
The PostgreSQL log under <PHANTOM_HOME>/data/db/pg_log folder to confirm whether WAL files are being archived successfully or the detailed error why archive command failed
The size and content of the pg_wal directory (<PHANTOM_HOME>/data/db/pg_wal/) to determine WAL backlog
Update the pgBackRest configuration file to increase the archive-timeout value:
Open the config file:
vi <PHANTOM_HOME>/etc/pgbackrest.conf
Locate the [phantom] stanza (or create one if not present) and add or modify the following:
[phantom]
pg1-path=/opt/phantom/data/db
archive-timeout=120
Save the file and retry:
<PHANTOM_HOME>/bin/phenv ibackup --setup
If increasing the timeout does not resolve the issue, you can bypass WAL archiving checks by temporarily disabling the pgBackRest-related configuration.
Open the PostgreSQL configuration:
vi <PHANTOM_HOME>/data/db/postgresql.conf
Comment out the following line:
#include_if_exists 'postgresql.ibackup.conf'
Restart PostgreSQL:
<PHANTOM_HOME>/bin/phsvc restart postgresql-11"
Once the WAL files are cleaned up, you can re-run ibackup setup again.
Please refer to this link for ibackup tools Splunk SOAR (On-premises) backup tools | Splunk Docs
1. Turned ibackup off: `ibackup --off`
2. Deleted all backups: `ibackup --delete-all`
3. Restarted Postgres: `phsvc restart postgresql`
4. You can re-run ibackup setup again
| Revision | Publish Date | Comments |
|---|---|---|
1.0 |
30-Sep-2026
|
Initial Release |