Vlad Khorsun
Vlad Khorsun
> In the case of TRIM with UNICODE_CI_AI, it converts the original string (expected to be in NFC) to NFD, remove accents and re-encode in NFC. That changes the length...
You may use expression index right now: ``` CREATE UNIQUE INDEX FOO_UQ_IDX ON FOO_TBL COMPUTED BY (CASE WHEN STOP_TS IS NULL THEN UNIQUE_VALUE END) ```
Run validation on regular basis (gfix -val -full) and check for messages at firebird.log put by validation
More information is in firebird.log. Warnings are harmless, don't worry. I never seen data loss because of sweep. So far there is not enough information to make any conclusion.
A lot of warnings on pointer pages is not good. Validation fixed it all, btw. It happens sooner of all after crash (or kill) of server process. To be safe,...
Usually such inconsistencies related with not correct stop of database without flushing of page cache. Check also forced writes setting on that db. I'll re-check careful writes logic around `swept`...
From description above it looks like you need select cu.cust_id ,cu.name, (select count(*) from invoice inv where inv.cust_id = cu.cust_id) as invoice_count from customer as cu Index on `invoice.cust_id` might...
What is the real problem ? Few hundreds of records ? Looks strange for me ;) Anyway, I see following possible ways to keep less records in RDB$BACKUP_HISTORY: - make...
> * make engine to allow to DELETE FROM RDB$BACKUP_HISTORY, probably by SYSDBA\Owner only, or > > * add command-line switch to the nbackup to avoid inserting record into RDB$BACKUP_HISTORY,...
Number of databases doesn't matters, as RDB$BACKUP_HISTORY is per-database. Even if one run nbackup every minute (which I highly doubt is possible on practice), then one will have 60 *...