ExportXMLWordPrintable

    • Type: Problem report
    • Resolution: Unresolved
    • Priority: Trivial
    • None
    • Affects Version/s: 7.4.16
    • Component/s: Templates (T)
    • None
    • Environment:
      Template: Veeam Backup & Replication by HTTP v 7.4-6
      Veeam Backup & Replication v 13.1.2.57 {$VEEAM.API.VERSION} 1.3-rev2

      ZBX-28099 keyed sessions on name and type, which unfortunately does not account for retries. Veeam records each retry as a new session and links it to the first attempt through originalSessionId, and a retry can carry a different name than the original, for example by appending "(Incremental)" or "(Full)" to the original job name, or in reverse the subsequent job might be the original job name without "(Incremental)". The main point is that the "name" field can't be trusted as a primary source of truth here.

      Evidence

      One job (same jobId on every row), GET /api/v1/sessions:

      creationTime      name                  result   id        originalSessionId
      2026-11-12 22:00  SomeJobName (Incremental)    Failed   892f64d7  892f64d7
      2026-11-13 09:20  SomeJobName                  Failed   fb3a7fa9  892f64d7   <- retry, different name
      2026-11-13 17:15  SomeJobName (Incremental)    Failed   da3e9a0b  da3e9a0b
      2026-11-13 17:30  SomeJobName (Incremental)    Success  5ac6bee0  da3e9a0b   <- retry, recovered
      

      Effect

      • The 17:15 failure raises an alert even though its retry succeeded 15 minutes later.
      • The item for the session with the name SomeJobName holds only Failed sessions, and a later success arrives under another session name, so the alert stays stuck until the sessions age out.
      • A session that ends in Warning never recovers the trigger.

      Expected

      Alert on the final attempt of a run: the latest session per originalSessionId, or the job-level lastResult from /jobs/states keyed on jobId. Both fields are in the list rows.

      Additionally, session truncation can lead to an invalid result.

      The sessions request has no limit, so it is truncated: in the above instance pagination shows a total of 278 sessions, but with the API (at least on my two VBR 13.1 Linux servers) having a default limit of 200 we end up missing sessions and getting data like so:

      "limit": 200, "count": 200, "total": 278
      

      For this I recommend setting up a pagination counter or adding something like "&limit=2000" to the API call, although the latter will eventually truncate sessions anyhow on a very busy system and can potentially lead to load issues - presumably Veeam set the default to 200 for a reason.

            Assignee:
            Petro Sklyar
            Reporter:
            Mike Ely
            Votes:
            0 Vote for this issue
            Watchers:
            1 Start watching this issue

              Created:
              Updated: