ServiceRadar v1.4.44
CSV device import upserts instead of skipping, copies tag pairs into
All Metadata, and SRQL actually returns those tags on in:devices
rows. A native-add-on seeder placeholder no longer blocks the signed
bundle. Downsampled series can split by a tag key.
Whats New
1.4.44
- Inventory: CSV import preview no longer CaseClauseError's on a
finished upload. A re-import upserts matching devices instead of
skipping them as already-exist, merges spreadsheet tags, and copies
key=valuepieces from the tags column into metadata so All Metadata
on device details is populated. Extra CSV columns still overlay the
same keys. (PRs #4040, #4042) - Inventory: a verified native-add-on import can replace an unverified
seeder placeholder for the same version, so first-party add-ons are
no longer frozen at the last pre-seeder bundle. (PR #4041) - SRQL:
in:devicesrows include thetagsJSONB map, matching the
filters and grouping that already usedtags.<key>. Dashboard
packages that declare a tags field no longer receive empty maps.
(PR #4044) - SRQL: downsampled timeseries can split with
series:tags.<key>, the
query shape fleet dashboards need thatstats: bycannot express
because it has no time buckets. (PR #4037) - CI: the release large-ingestion waiter installs
ghon the signing
runner so the wait is a real gate instead of FileNotFoundError.
(PR #4036)
Upgrade Notes
- No schema migrations in this release.
migrations.expectedVersion
stays20260825031000. - Re-import a device CSV to fill All Metadata on rows that landed
before 1.4.44. Tags were already stored; metadata was not. - Native add-on seeder/import: after upgrade, a signed bundle for a
version that only had a seeder placeholder can replace that row.
Netprobe and other first-party add-ons stuck on a pre-seeder version
become importable. - Dashboard packages that group on
tags.<key>start working against
in:deviceswithout a metadata fallback. Republish is not required
if the package already requested thetagsfield.