Savings timeline stills crashes SQL server

Client: Resolve 17b6 on Windows10 w/ latest updates
Server: Resolve 17b6 on Mac Mini M1, OS X 11.1

When we save multiple gallery stills to the server ("Grab all stills" to PowerGrades), it seems to overwhelm and crash it. The PostGreSQL log actually contains some of the stills' binary bitmap data (probably MIME-encoded), or so it seems...
Here are relevant snippets. Full logs attached.

Code: Select all

2021-01-26 08:24:02 PST STATEMENT:  DELETE FROM "Gallery::GyStillRef" WHERE "Gallery::GyStillRef_id" IN ('51c1a7da-f4b0-4b3f-b2d3-0dd134d943e1')
2021-01-26 08:24:02 PST ERROR:  current transaction is aborted
2021-01-26 08:24:03 PST STATEMENT:  DELETE FROM "Gallery::GyStillRef" WHERE "Gallery::GyStillRef_id" IN ('8a489e1e-f301-4115-87ca-d653e9daf106')
2021-01-26 08:24:04 PST ERROR:  index "Gallery::GyStill_pClipFullVer_idx" contains unexpected zero page at block 6
2021-01-26 08:24:04 PST HINT:  Please REINDEX it.
2021-01-26 08:24:04 PST CONTEXT:  SQL statement "SELECT 1 FROM ONLY "public"."Gallery::GyStill" x WHERE $1 OPERATOR(pg_catalog.=) "pClipFullVer" FOR KEY SHARE OF x"
2021-01-26 08:24:04 PST STATEMENT:  COMMIT
2021-01-26 08:28:12 PST ERROR:  index "Gallery::GyStill_pClipFullVer_idx" contains unexpected zero page at block 6
2021-01-26 08:28:12 PST HINT:  Please REINDEX it.
2021-01-26 08:28:12 PST STATEMENT:  INSERT INTO "Gallery::GyStill" ("Gallery::GyStill_id", "DbType", "FieldsBlob", "SrcHint", "SrcType", "GalleryPath", "Label", "RecTC", "SrcTC", "DpxDescriptor", "Width", "Height", "BitDepth", "PAR", "Endianship", "CreateTime", "pClipFullVer", "pTrackVer", "PrimaryCCMode", "Vsr", "ClipThumbnails", "TrackThumbnails", "ReelName") VALUES ('98c9c15c-5e08-47b3-8bb7-25b38e15fac3', 'Gallery::GyStill', '\x00000001 [....lots of MIME here........] 'TI31009N11_N011C001_201010SI')
2021-01-26 08:29:49 PST WARNING:  hash indexes are not WAL-logged and their use is discouraged
2021-01-26 08:29:49 PST CONTEXT:  SQL statement "CREATE INDEX "BtLockableBlobMap_SM_User_id_idx" ON "BtLockableBlobMap" USING hash ("SM_User_id")"
   PL/pgSQL function create_index_for_foreign_keys() line 6 at EXECUTE
2021-01-26 08:29:49 PST WARNING:  hash indexes are not WAL-logged and their use is discouraged
2021-01-26 08:29:49 PST CONTEXT:  SQL statement "CREATE INDEX "Sm2RenderCacheBlob_TiItemOwner_idx" ON "Sm2RenderCacheBlob" USING hash ("TiItemOwner")"
   PL/pgSQL function create_index_for_foreign_keys() line 6 at EXECUTE
2021-01-26 08:30:09 PST LOG:  could not receive data from client: Connection reset by peer

[Manually rebooted the machine]

2021-01-26 09:36:23 PST LOG:  database system was interrupted; last known up at 2021-01-26 09:20:44 PST
2021-01-26 09:36:23 PST LOG:  database system was not properly shut down; automatic recovery in progress
2021-01-26 09:36:23 PST LOG:  redo starts at 5/5569AA90
2021-01-26 09:36:23 PST LOG:  invalid record length at 5/5766C740
2021-01-26 09:36:23 PST LOG:  redo done at 5/5766C6D0
2021-01-26 09:36:23 PST LOG:  last completed transaction was at log time 2021-01-26 09:19:39.782236-08
2021-01-26 09:36:24 PST LOG:  MultiXact member wraparound protections are now enabled
2021-01-26 09:36:24 PST LOG:  autovacuum launcher started
2021-01-26 09:36:24 PST LOG:  database system is ready to accept connections


Of course maybe this got fixed in the M1-compatible 17.1b7, maybe not. I'll check.

Still happens with 17.1b8.

[Dwaine Maggart]

Do you have Resolve Project Server on the Mac Mini? If so, launch it, select the database in question, and select the Optimize icon.