cliu/ci-server-engineer
1
0
Fork 0
Code Issues Pull requests Projects Releases Packages Wiki Activity Actions

Record the classifier storage cache test (ISS-050)

Browse source 
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
This commit is contained in:
Clio Liu 2026-10-01 17:32:47 +10:00
parent f016f516b8
commit 5b6a936592
3 changed files with 6 additions and 5 deletions
Show stats Download patch file Download diff file Expand all files Collapse all files

5
.claude/skills/audit-trail/tasks/configure-audit.md
View file

@ -26,8 +26,9 @@ NAME,STORAGE
`dssqld -m`, or in the Engineering Module *Configuration › Audit trail › Event classifiers* (the change itself is audited: `AUDIT_EVENT_CLAS_DF` `STORAGE` `MODIFY`). `dssqld -m`, or in the Engineering Module *Configuration › Audit trail › Event classifiers* (the change itself is audited: `AUDIT_EVENT_CLAS_DF` `STORAGE` `MODIFY`).
**Not immediate:** the processes that report events (e.g. `DSSSDA`) cache the storage flags and **Not immediate:** the processes that report events (e.g. `DSSSDA`) cache the storage flags and
re-read them every `ECL_REFRESH` seconds (default 1800). In the test a switched-off classifier was re-read them every `ECL_REFRESH` seconds (default 1800). Verified: a classifier switched off was
still recorded right after the change (ISS-050). Confirm with the user before still recorded right after the change and no longer 31 minutes later; switching it back on is
delayed the same way. Allow up to 30 minutes or restart CI Server (ISS-050). Confirm with the user before
switching anything off. switching anything off.
### Item option *Audit trailing wanted* (`AUDIT_INFO`) ### Item option *Audit trailing wanted* (`AUDIT_INFO`)

2
knowledge/issues.md
View file

@ -69,5 +69,5 @@ development system (see "Test system" in `progress.md`); they do not exist on ot
| ISS-047 | 2026-10-01 | REPORT (RQL manuals) | The manuals disagree: default `@WORKSPACE` 500 KB (IM 36K10R40 §2.2.2) vs 16 KB (IM 36K10C30 table 2.1-1, IM 36K10P20 §9.6); maximum `@WIDTH` 512 (R40, P20 §4) vs 5000 (P20 §9.6); `@FORMFEED` described with opposite meanings (C30: not 1 = form feed; P20: 0 = blank lines, default 1). Installed set-up files set `@workspace` explicitly (500 in `rptge1.rpt`, 100 in `rptgen.rpt`) and `@formfeed 0` for the interactive generator. | open | Not tested (no printer); the set-up files make the defaults irrelevant in practice. | `knowledge/manuals/IM36K10R40-01EN.md` §2; `knowledge/manuals/IM36K10P20-01EN.md` §3 | | ISS-047 | 2026-10-01 | REPORT (RQL manuals) | The manuals disagree: default `@WORKSPACE` 500 KB (IM 36K10R40 §2.2.2) vs 16 KB (IM 36K10C30 table 2.1-1, IM 36K10P20 §9.6); maximum `@WIDTH` 512 (R40, P20 §4) vs 5000 (P20 §9.6); `@FORMFEED` described with opposite meanings (C30: not 1 = form feed; P20: 0 = blank lines, default 1). Installed set-up files set `@workspace` explicitly (500 in `rptge1.rpt`, 100 in `rptgen.rpt`) and `@formfeed 0` for the interactive generator. | open | Not tested (no printer); the set-up files make the defaults irrelevant in practice. | `knowledge/manuals/IM36K10R40-01EN.md` §2; `knowledge/manuals/IM36K10P20-01EN.md` §3 |
| ISS-048 | 2026-10-01 | REPORT (ad-hoc parameters) | The prompt `[ITEM:NAME like "CI-Eng-Sample.TANK1.*"]Item for statistics:` (filter form as in IM 36K10C30 §2.1) gives an item tree browser in the Engineering Module generate dialog, but the tree shows all sections, not only `TANK1` items (checked by the user). | open | Try other filter forms (`%` wildcard, `SECTION = …`, `[ITEM:TAG="…"]`) and the display report component. | `.claude/skills/reporting/tasks/define-reports.md`; `knowledge/manuals/IM36K10C30-01EN.md` §2; `examples/CI-Eng-Sample/18_report_df.qli` comment | | ISS-048 | 2026-10-01 | REPORT (ad-hoc parameters) | The prompt `[ITEM:NAME like "CI-Eng-Sample.TANK1.*"]Item for statistics:` (filter form as in IM 36K10C30 §2.1) gives an item tree browser in the Engineering Module generate dialog, but the tree shows all sections, not only `TANK1` items (checked by the user). | open | Try other filter forms (`%` wildcard, `SECTION = …`, `[ITEM:TAG="…"]`) and the display report component. | `.claude/skills/reporting/tasks/define-reports.md`; `knowledge/manuals/IM36K10C30-01EN.md` §2; `examples/CI-Eng-Sample/18_report_df.qli` comment |
| ISS-049 | 2026-10-01 | AUDIT (item option *Audit trailing wanted*) | IM 36K10G10 ch. 9 and the classifier list (IM 36K10L25 App. 1, `ITEM.VALUE.MODIFY.<source>`, `ITEM.HIGH-LIMIT…`) suggest item changes are audited per item when `AUDIT_INFO` = 1. On R1.03.00 none of the tested tools produced an `ITEM.*` event: `itmval -p` writes are not audited at all (with `AUDIT_INFO` 0 or 1); value writes through DSS (`dssqld -m ITEM_VAL`) and Engineering Module *Update value* are audited as dataset events (`ITEM_VAL`, `ITEM_VALUE` / `ITEM_VALUE_STR` / `STATUS`) for every item regardless of `AUDIT_INFO`. Operator-display writes and PROCESS writes (`opc.sup` `AUDIT = YES`, `COMPUTE` classifiers off by default) were not tested (no input field in the sample displays). | open | Test an operator value input on an item with `AUDIT_INFO` 1 (needs a display with an input field) and, if wanted, PROCESS with `AUDIT = YES`. | `.claude/skills/audit-trail/tasks/configure-audit.md` §1; `.claude/skills/audit-trail/tasks/read-audit-trail.md` §2; `knowledge/manuals/IM36K10L25-01EN.md` §3 | | ISS-049 | 2026-10-01 | AUDIT (item option *Audit trailing wanted*) | IM 36K10G10 ch. 9 and the classifier list (IM 36K10L25 App. 1, `ITEM.VALUE.MODIFY.<source>`, `ITEM.HIGH-LIMIT…`) suggest item changes are audited per item when `AUDIT_INFO` = 1. On R1.03.00 none of the tested tools produced an `ITEM.*` event: `itmval -p` writes are not audited at all (with `AUDIT_INFO` 0 or 1); value writes through DSS (`dssqld -m ITEM_VAL`) and Engineering Module *Update value* are audited as dataset events (`ITEM_VAL`, `ITEM_VALUE` / `ITEM_VALUE_STR` / `STATUS`) for every item regardless of `AUDIT_INFO`. Operator-display writes and PROCESS writes (`opc.sup` `AUDIT = YES`, `COMPUTE` classifiers off by default) were not tested (no input field in the sample displays). | open | Test an operator value input on an item with `AUDIT_INFO` 1 (needs a display with an input field) and, if wanted, PROCESS with `AUDIT = YES`. | `.claude/skills/audit-trail/tasks/configure-audit.md` §1; `.claude/skills/audit-trail/tasks/read-audit-trail.md` §2; `knowledge/manuals/IM36K10L25-01EN.md` §3 |
| ISS-050 | 2026-10-01 | AUDIT (classifier `STORAGE`) | Setting `AUDIT_EVENT_CLAS_DF.STORAGE` = 0 for `DATASET.NULL-ATTR.MODIFY.ENT-ENG-MOD` at 17:01:12 did not stop the next `dssqld -m ITEM_VAL` change from being stored (17:01:12.573). IM 36K10L25 §3.2 explains a per-process cache refreshed every `ECL_REFRESH` s (default 1800); the notifier here is `DSSSDA`, so the change should apply within 30 minutes. | investigating | Re-test after 17:31 on 2026-10-01, then set `STORAGE` back to 1. | `.claude/skills/audit-trail/tasks/configure-audit.md` §1; `knowledge/manuals/IM36K10L25-01EN.md` §3 | | ISS-050 | 2026-10-01 | AUDIT (classifier `STORAGE`) | Setting `AUDIT_EVENT_CLAS_DF.STORAGE` = 0 for `DATASET.NULL-ATTR.MODIFY.ENT-ENG-MOD` at 17:01:12 did not stop the next `dssqld -m ITEM_VAL` change from being stored (17:01:12.573). IM 36K10L25 §3.2 explains a per-process cache refreshed every `ECL_REFRESH` s (default 1800); the notifier here is `DSSSDA`, so the change should apply within 30 minutes. | open | Re-tested: at 17:32:18 (after the 30-minute refresh) the same change was **not** stored, so the flag works with a delay. Set back to 1 at 17:32; the next change was still not stored (the cache works in both directions). Documented: allow up to `ECL_REFRESH` (1800 s) or restart CI Server. Close when the user confirms. | `.claude/skills/audit-trail/tasks/configure-audit.md` §1; `knowledge/manuals/IM36K10L25-01EN.md` §3 |
| ISS-051 | 2026-10-01 | AUDIT (manual vs installed) | Differences between IM 36K10L25 and R1.03.00: (1) App. 1 lists subject `F/T-SYSTEM`, the installed classifiers use `CI-SERVER` (logins) and also `EAL OPERATION` (subject and action) and source `VIA-OPC-UAS`, not in the manual; 463 classifiers installed. (2) App. 1 explanations of `SHELVE` and `UN-SHELVE` are swapped. (3) `QUESIZE`: manual default and maximum 200 kB, `adtsys.sup` help default 10 kB, minimum 10 (file sets 200). (4) `TIMEFMT` default: manual `DD/MM/YY hh:mm:ss.ppp`, `adtsys.sup` help without `.ppp`; `ADTLOG` output shows milliseconds (manual right). (5) `dssqld` changes are recorded with source `ENT-ENG-MOD`, not `QUICKLOAD`. (6) The selection expression language is not described in the manual; the compiler listing gives the words and operators (no `LIKE`). (7) `ADT` group lifetime 52 weeks on the test system vs 2 days default when `ADTSTO` creates it (§2.1.1). | open | Documented in the L25 summary and the `audit-trail` skill. Close when the user confirms. | `knowledge/manuals/IM36K10L25-01EN.md`; `.claude/skills/audit-trail/tasks/*.md` | | ISS-051 | 2026-10-01 | AUDIT (manual vs installed) | Differences between IM 36K10L25 and R1.03.00: (1) App. 1 lists subject `F/T-SYSTEM`, the installed classifiers use `CI-SERVER` (logins) and also `EAL OPERATION` (subject and action) and source `VIA-OPC-UAS`, not in the manual; 463 classifiers installed. (2) App. 1 explanations of `SHELVE` and `UN-SHELVE` are swapped. (3) `QUESIZE`: manual default and maximum 200 kB, `adtsys.sup` help default 10 kB, minimum 10 (file sets 200). (4) `TIMEFMT` default: manual `DD/MM/YY hh:mm:ss.ppp`, `adtsys.sup` help without `.ppp`; `ADTLOG` output shows milliseconds (manual right). (5) `dssqld` changes are recorded with source `ENT-ENG-MOD`, not `QUICKLOAD`. (6) The selection expression language is not described in the manual; the compiler listing gives the words and operators (no `LIKE`). (7) `ADT` group lifetime 52 weeks on the test system vs 2 days default when `ADTSTO` creates it (§2.1.1). | open | Documented in the L25 summary and the `audit-trail` skill. Close when the user confirms. | `knowledge/manuals/IM36K10L25-01EN.md`; `.claude/skills/audit-trail/tasks/*.md` |

4
knowledge/manuals/IM36K10L25-01EN.md
View file

@ -82,8 +82,8 @@ as *dataset* events — `SUBJECT_GROUP` = dataset name (`ITEM_DF`, `ITEM_VAL`, `
`CI-SERVER` `LOGIN`/`LOGOUT` pair. `itmval -p` writes are not audited. The item option `CI-SERVER` `LOGIN`/`LOGOUT` pair. `itmval -p` writes are not audited. The item option
`AUDIT_INFO` (*Audit trailing wanted*, default 0; also on `SUB_ITEM_DF`, `EXTERNAL_ALARM_DF`) made `AUDIT_INFO` (*Audit trailing wanted*, default 0; also on `SUB_ITEM_DF`, `EXTERNAL_ALARM_DF`) made
no difference for these tools; it is meant for the `ITEM.*` classifiers from operator/PROCESS no difference for these tools; it is meant for the `ITEM.*` classifiers from operator/PROCESS
writes (not tested, ISS-049). A changed `STORAGE` flag is cached by the notifying process writes (not tested, ISS-049). A changed `STORAGE` flag takes effect only after the notifying process
(`ECL_REFRESH`, ISS-050). re-reads it (`ECL_REFRESH`, up to 30 minutes, both directions; verified, ISS-050).
## 4. Setup files (§4.2) ## 4. Setup files (§4.2)