Symptoms: Backup Job Failing, iSCSI Target Unreachable, Fix Is VLAN Config¶
Domains: linux_ops | datacenter_ops | networking Level: L3 Estimated time: 45 min
Initial Alert¶
Backup monitoring system fires at 02:30 UTC:
CRITICAL: Nightly backup FAILED — db-primary-01
Job: pg-backup-nightly
Duration: 0s (expected: 15-20 min)
Error: "pg_dump: could not open output file /mnt/backup/pg_backup_20260319.sql.gz: No such file or directory"
Follow-up alert:
CRITICAL: Mount point /mnt/backup is not mounted on db-primary-01
WARNING: Backup SLA breach — 2 consecutive nightly failures
Observable Symptoms¶
- The backup job fails because
/mnt/backupis not mounted. mount | grep backupon db-primary-01 returns nothing — the filesystem is not mounted./etc/fstabshows the backup mount as an iSCSI target:iscsi-san.storage.internal:/backup-lun-04 /mnt/backup ext4 _netdev,defaults 0 0.iscsiadm -m sessionshows no active iSCSI sessions.ping iscsi-san.storage.internalsucceeds (ICMP echo works).- The backup worked fine 3 days ago. Nothing changed on the database server.
- Two other database servers (db-primary-02, db-replica-01) are also failing to mount their iSCSI backup volumes.
The Misleading Signal¶
A missing mount point with a backup failure looks like a Linux filesystem issue — maybe the iSCSI initiator service crashed, fstab is corrupted, or the mount was accidentally unmounted. The fact that ping to the storage array works makes it look like the storage target is up but the iSCSI service on the Linux host is broken. Engineers start investigating iscsid, checking iSCSI target discovery, and reviewing systemd mount units.