---
manual: IM 36K10L36-01EN
title: ITEM Service Engineering - ITM
edition: 3rd Edition, Sep. 2023
version: R1.03
status: from docs; itm.sup, item_start.cmd and itmdbg statistics checked on the test system (2026-09-30); itmval -g/-p verified in phase 1; itmval -f crashes (ISS-022)
---

# ITEM Service Engineering - ITM (R1.03)

ITM is the real-time item database (item common, shared memory). Section numbers refer to the
manual.

## 1. Processes (§1.1–1.2)

| Process | Role |
|---|---|
| `ITM` | item handler, message interface |
| `ITMTIM` | sends buffered item events; recalculates integrators |
| `ITMCPM` / `ITMCPS` | item copy master / slave for distributed items between nodes (buffered, compressed; after a lost MDUR link it rescans all or only changed master items in limited batches). Restart `ITMCPM` after deleting a distributed group. |
| `ITMLIMUPD` | copies limit/deadband changes to other nodes |
| `ITMREM` | cleans up event requests of processes that disappeared |
| `itmval` | command-line get/put |

`itmtim [<period ms, default 100>] [<integration step ms, default 1000>] [<save count>]`:
save count = integrators saved to disk per recalculation. **Installed** `item_start.cmd`
starts `itmtim 100 1000 1 200 1` (five arguments; the last two are not documented). Also
started: `itmaah`, `intitmini`, `itmlimupd`, `itmrem`.

## 2. `itm.sup` (§1.3)

| Setting (keyword) | Manual default | Test system |
|---|---|---|
| Queue size (`queue_size`) | 200 KB | 200 |
| Save items (`item_save`): ITEM_BASE (per item's *remember last value*), ALWAYS, NEVER | ITEM_BASE | ITEM_BASE |
| Max item groups (`max_groups`) | 255 | 255 |
| Max items per group (`max_number`) | 65534 (255 × 65534 = 16,711,170 items) | 65534 |
| Max application info per item (`max_appl_info`) | ≤ 20 bytes always fits | 0 |
| Max update info per item (`max_upd_info`) | 512 | 512 |
| Item pool size (`pool_kbytes`), guide `0.07 × groups × items + appl KB × groups × items + 10` | **50000 KB** | **650000** |
| Integration granule (`itm_int_secs`), system-wide integrator time unit | 1 s | 1 (60 tested, ISS-010) |
| Lock timeout (`lock_timeout_secs`) | 60 s | 60 |
| Save to disk (`disk_save`) | YES | yes |
| Not regained (`not_regained`): after start, mark possibly stale values until the first update; hidden in trends | not described | no (default block) |

## 3. `itmval` (§1.4)

```
itmval -g [-l] -n <item>
itmval -p [-l] -n <item> [-v <value>] [-s <status>] [-q <quality>] [-o] [-t [<date>] <time> [GMT|LST]]
itmval -f <file> <vsqo> [-l] -n <item>
```

`-l` = use this node instead of the host; `-o` = overwrite a blocked item; `-t` = update time
(default local time). `-f` should append an `itmval -p …` restore line with the chosen parts
(value, status, quality, options) to a command file, but on the test system it **crashes**
(segmentation fault / access violation) and writes nothing (ISS-022). The manual's
space-separated name form (`-n <inst> <unit> <tag>`) is rejected ("no (extra) command line
arguments allowed"); use the dotted full name. Practiced: skill `system-management`, task
`read-write-item.md`.

## 4. `itmdbg` (§2)

Start `itmdbg`, then a code; answers can be piped, e.g. read-only statistics:
`printf '2\n-1\n' | itmdbg`.

| Code | Shows / does |
|---|---|
| 1 | main info (root pointers, criterion and event request bits) |
| 2 | statistics: common size/free/fragments, configured limits, **nr of items**, integration secs, save mode, counters (item/sub-item updates, gets, buffered events, automatic saves, flow control) since system start and since debugger start |
| 3 / 4 | item / sub-item info by group and sequence number (value, status, options, limits, update times, name) |
| 5–11 | action blocks, main actions, task table (event requesters), event buffers, mask info (criterion bits: 1 FIRST, 2 EACH, 4 VAL, 8 STA, 10 DEL, …), item pointers, last update |
| 12 | unlock item data area (after a killed process; if pointers were being changed, restart ITEM Service) |
| 13 | save all item info to disk (before a shutdown, for a clean warm restart) |
| 14, 15, 17, 18 | flow control, integrator list, locked items, group–node relation |
| 16 | **delete item** from the in-core database (test/repair only) |
| 19 / 20 | most frequently updated items (10–1000) / reset those counters |

Test system (`itmdbg` code 2): common 665,600,000 bytes, free 665,307,508, 203 items,
integration 1 s, save mode item base. The same numbers appear in the automatic snapshots
`smd_*_itmdbg_all.log` (skill `system-management`, task `tuning-checks.md`).