Overview
Comment: | New logic on delaying when can't run more tests. One more case of dbstruct used inside transaction found and fixed. |
---|---|
Downloads: | Tarball | ZIP archive | SQL archive |
Timelines: | family | ancestors | descendants | both | v1.80 |
Files: | files | file ages | folders |
SHA1: |
a67d8dbadb8879b5657763380370123b |
User & Date: | matt on 2023-01-13 05:07:06 |
Other Links: | branch diff | manifest | tags |
Context
2023-01-13
| ||
08:22 | Yet another db in transaction problem fixed. check-in: 95b6039c92 user: matt tags: v1.80 | |
05:07 | New logic on delaying when can't run more tests. One more case of dbstruct used inside transaction found and fixed. check-in: a67d8dbadb user: matt tags: v1.80 | |
2023-01-12
| ||
21:49 | Calling db:with-db inside a transaction was bad. Disallowed calling db:with-db with db and factored out few calls to take db. TCP connections seems stable. Fingers crossed. check-in: fb48802dd5 user: matt tags: v1.80 | |
Changes
Modified db.scm from [3ad6d284e8] to [8baf166ad8].
︙ | ︙ | |||
3309 3310 3311 3312 3313 3314 3315 | (db:with-db dbstruct run-id #f (lambda (dbdat db) (let ((tr-res (sqlite3:with-transaction db (lambda () | | | < < < | | | | | | < > > > > > | 3309 3310 3311 3312 3313 3314 3315 3316 3317 3318 3319 3320 3321 3322 3323 3324 3325 3326 3327 3328 3329 3330 3331 3332 3333 3334 3335 3336 3337 3338 3339 3340 3341 3342 3343 3344 | (db:with-db dbstruct run-id #f (lambda (dbdat db) (let ((tr-res (sqlite3:with-transaction db (lambda () (let* ((state-status-counts (db:get-all-state-status-counts-for-run-db db run-id)) (state-statuses (db:roll-up-rules state-status-counts #f #f )) (newstate (car state-statuses)) (newstatus (cadr state-statuses))) (if (or (not (eq? newstate curr-state)) (not (eq? newstatus curr-status))) (db:set-run-state-status db run-id newstate newstatus ))))))) (mutex-unlock! *db-transaction-mutex*) tr-res)))) (define (db:get-all-state-status-counts-for-run-db db run-id) (sqlite3:map-row (lambda (state status count) (make-dbr:counts state: state status: status count: count)) db "SELECT state,status,count(id) FROM tests WHERE run_id=? GROUP BY state,status;" run-id )) (define (db:get-all-state-status-counts-for-run dbstruct run-id) (db:with-db dbstruct #f #f (lambda (dbdat db) (db:get-all-state-status-counts-for-run-db dbstruct run-id)))) ;; BBnote: db:get-all-state-status-counts-for-test returns dbr:counts object aggregating state and status of items of a given test, *not including rollup state/status* ;; ;; NOTE: This is called within a transaction ;; (define (db:get-all-state-status-counts-for-test db run-id test-name item-path item-state-in item-status-in) (let* ((test-info (db:get-test-info-db db run-id test-name item-path)) |
︙ | ︙ |
Modified runs.scm from [0255133584] to [dd48ee309d].
︙ | ︙ | |||
1277 1278 1279 1280 1281 1282 1283 | ", ")) (thread-sleep! 0.051) (list hed tal reg reruns)) ;; If no resources are available just kill time and loop again ;; ((not have-resources) ;; simply try again after waiting a second | | > | > > > > > > > | 1277 1278 1279 1280 1281 1282 1283 1284 1285 1286 1287 1288 1289 1290 1291 1292 1293 1294 1295 1296 1297 1298 1299 1300 1301 1302 1303 | ", ")) (thread-sleep! 0.051) (list hed tal reg reruns)) ;; If no resources are available just kill time and loop again ;; ((not have-resources) ;; simply try again after waiting a second (if (runs:lownoise "no resources" 600) (debug:print-info 1 *default-log-port* "no resources to run new tests, waiting ...")) ;; Have gone back and forth on this but db starvation is an issue. ;; wait one second before looking again to run jobs. ;; (thread-sleep! 0.25) ;; new logic. ;; If it has been more than 10 seconds since we were last here don't wait at all ;; otherwise sleep 2 seconds to give db a rest and let dashboard read data (if (runs:lownoise "frequent-no-resources" 10) (thread-sleep! 0.25) ;; no significant delay (thread-sleep! 2)) ;; could have done hed tal here but doing car/cdr of newtal to rotate tests (list (car newtal)(cdr newtal) reg reruns)) ;; This is the final stage, everything is in place so launch the test ;; ((and have-resources (or (null? prereqs-not-met) |
︙ | ︙ |