Differences From Artifact [91ffe1108a]:
- File rmt.scm — part of check-in [08646db430] at 2022-11-22 06:36:06 on branch v1.70-nohomehost — Remove case 6 from rmt:send-receive. This was pinging the server constantly and opening addtional connections that don't appear to being closed. Also, just let the connection die, it should retry. (user: matt, size: 52006) [annotate] [blame] [check-ins using] [more...]
To Artifact [ef90d475fb]:
- File
rmt.scm
— part of check-in
[67b18cc991]
at
2023-01-16 10:00:35
on branch v1.80-refactor
— Major refactor/clean up of client connection stuff. Removed old crufty approach of having both client and server side calls in http-transport. Plan would be to eventually get rid of http-transport.
Open files slowly grows. bad argument type in http-transport. Line 285 (close-connection! fullurl) is wrong. (user: matt, size: 49946) [annotate] [blame] [check-ins using] [more...]
︙ | ︙ | |||
40 41 42 43 44 45 46 | ;;====================================================================== ;; S U P P O R T F U N C T I O N S ;;====================================================================== ;; if a server is either running or in the process of starting call client:setup ;; else return #f to let the calling proc know that there is no server available ;; | | < | | | 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60 61 | ;;====================================================================== ;; S U P P O R T F U N C T I O N S ;;====================================================================== ;; if a server is either running or in the process of starting call client:setup ;; else return #f to let the calling proc know that there is no server available ;; (define (rmt:get-connection-info areapath runremote) ;; TODO: push areapath down. (let* ((cinfo (if (remote? runremote) (remote-conndat runremote) #f))) (if cinfo cinfo (if (server:check-if-running areapath) (client:setup areapath runremote) #f)))) (define (rmt:on-homehost? runremote) (let* ((hh-dat (remote-hh-dat runremote))) (if (pair? hh-dat) (cdr hh-dat) (begin |
︙ | ︙ | |||
168 169 170 171 172 173 174 | ;; also, the expire-time calculation might not be correct. We want, time-since-last-server-access > (server:get-timeout) ;; ;;DOT CASE4 [label="reset\nconnection"]; ;;DOT MUTEXLOCK -> CASE4 [label="have connection,\nlast_access > expire_time"]; {rank=same "case 4" CASE4} ;;DOT CASE4 -> "rmt:send-receive"; ;; reset the connection if it has been unused too long ((and runremote | | | > > | < | 167 168 169 170 171 172 173 174 175 176 177 178 179 180 181 182 183 184 185 186 187 188 | ;; also, the expire-time calculation might not be correct. We want, time-since-last-server-access > (server:get-timeout) ;; ;;DOT CASE4 [label="reset\nconnection"]; ;;DOT MUTEXLOCK -> CASE4 [label="have connection,\nlast_access > expire_time"]; {rank=same "case 4" CASE4} ;;DOT CASE4 -> "rmt:send-receive"; ;; reset the connection if it has been unused too long ((and runremote ;; (remote-conndat runremote) (> (current-seconds) ;; if it has been more than server-timeout seconds since last contact, close this connection and start a new on (+ (remote-last-access runremote) (remote-server-timeout runremote)))) (debug:print-info 0 *default-log-port* "Connection to " (remote-server-url runremote) " expired due to no accesses, forcing new connection.") (http-transport:close-connections runremote) ;; moving this setting of runremote conndat to #f to inside the http-transport:close-connections ;; (remote-conndat-set! runremote #f) ;; invalidate the connection, thus forcing a new connection. (mutex-unlock! *rmt-mutex*) (rmt:send-receive cmd rid params attemptnum: attemptnum)) ;;DOT CASE5 [label="local\nread"]; ;;DOT MUTEXLOCK -> CASE5 [label="server not required,\non homehost,\nread-only query"]; {rank=same "case 5" CASE5}; ;;DOT CASE5 -> "rmt:open-qry-close-locally"; |
︙ | ︙ | |||
196 197 198 199 200 201 202 | ;;DOT CASE6 [label="init\nremote"]; ;;DOT MUTEXLOCK -> CASE6 [label="on homehost,\nwrite query,\nhave server,\ncan't reach it"]; {rank=same "case 6" CASE6}; ;;DOT CASE6 -> "rmt:send-receive"; ;; on homehost and this is a write, we already have a server, but server has died ;; reinstate this keep-alive section but inject a time condition into the (add ... | | | | | | | | | | | | | | | | | | | 196 197 198 199 200 201 202 203 204 205 206 207 208 209 210 211 212 213 214 215 216 217 218 219 220 221 222 223 224 225 226 | ;;DOT CASE6 [label="init\nremote"]; ;;DOT MUTEXLOCK -> CASE6 [label="on homehost,\nwrite query,\nhave server,\ncan't reach it"]; {rank=same "case 6" CASE6}; ;;DOT CASE6 -> "rmt:send-receive"; ;; on homehost and this is a write, we already have a server, but server has died ;; reinstate this keep-alive section but inject a time condition into the (add ... ;; ;; ((and (cdr (remote-hh-dat runremote)) ;; on homehost ;; (not (member cmd api:read-only-queries)) ;; this is a write ;; (remote-server-url runremote) ;; have a server ;; (not (server:ping (remote-server-url runremote) (remote-server-id runremote)))) ;; server has died. NOTE: this is not a cheap call! Need better approach. ;; (debug:print 0 *default-log-port* "WARNING: server appears to have died, trying to reconnect, case 6") ;; (http-transport:close-connections area-dat: runremote) ;; make sure to clean up ;; (set! *runremote* (make-remote)) ;; (let* ((server-info (remote-server-info *runremote*))) ;; (if server-info ;; (begin ;; (remote-server-url-set! *runremote* (server:record->url server-info)) ;; (remote-server-id-set! *runremote* (server:record->id server-info))))) ;; (remote-force-server-set! runremote (common:force-server?)) ;; (mutex-unlock! *rmt-mutex*) ;; (debug:print-info 12 *default-log-port* "rmt:send-receive, case 6") ;; (rmt:send-receive cmd rid params attemptnum: attemptnum)) ;;DOT CASE7 [label="homehost\nwrite"]; ;;DOT MUTEXLOCK -> CASE7 [label="server not required,\non homehost,\na write,\nhave a server"]; {rank=same "case 7" CASE7}; ;;DOT CASE7 -> "rmt:open-qry-close-locally"; ;; on homehost and this is a write, we already have a server ((and (not (remote-force-server runremote)) ;; honor forced use of server, i.e. server NOT required (cdr (remote-hh-dat runremote)) ;; on homehost |
︙ | ︙ | |||
259 260 261 262 263 264 265 | (not (remote-conndat runremote))) (and (not (cdr (remote-hh-dat runremote))) ;; not on a homehost (not (remote-conndat runremote)))) ;; and no connection (debug:print-info 12 *default-log-port* "rmt:send-receive, case 9, hh-dat: " (remote-hh-dat runremote) " conndat: " (remote-conndat runremote)) (mutex-unlock! *rmt-mutex*) (if (not (server:check-if-running *toppath*)) ;; who knows, maybe one has started up? (server:start-and-wait *toppath*)) | | < < < < < < < < < < < < < < < < < < < < < < < | | | | | | | | | > | | | < < < | < | | 259 260 261 262 263 264 265 266 267 268 269 270 271 272 273 274 275 276 277 278 279 280 281 282 283 284 285 286 287 288 289 290 291 292 293 294 295 296 297 298 299 300 301 302 303 304 305 306 307 308 309 310 311 312 313 314 315 316 317 318 319 320 321 322 323 324 325 326 | (not (remote-conndat runremote))) (and (not (cdr (remote-hh-dat runremote))) ;; not on a homehost (not (remote-conndat runremote)))) ;; and no connection (debug:print-info 12 *default-log-port* "rmt:send-receive, case 9, hh-dat: " (remote-hh-dat runremote) " conndat: " (remote-conndat runremote)) (mutex-unlock! *rmt-mutex*) (if (not (server:check-if-running *toppath*)) ;; who knows, maybe one has started up? (server:start-and-wait *toppath*)) (remote-conndat-set! runremote (rmt:get-connection-info *toppath* runremote)) ;; calls client:setup which calls client:setup-http (rmt:send-receive cmd rid params attemptnum: attemptnum)) ;; TODO: add back-off timeout as ;;DOT CASE10 [label="on homehost"]; ;;DOT MUTEXLOCK -> CASE10 [label="server not required,\non homehost"]; {rank=same "case 10" CASE10}; ;;DOT CASE10 -> "rmt:open-qry-close-locally"; ;; all set up if get this far, dispatch the query ((and (not (remote-force-server runremote)) (cdr (remote-hh-dat runremote))) ;; we are on homehost (mutex-unlock! *rmt-mutex*) (debug:print-info 12 *default-log-port* "rmt:send-receive, case 10") (rmt:open-qry-close-locally cmd (if rid rid 0) params)) ;;DOT CASE11 [label="send_receive"]; ;;DOT MUTEXLOCK -> CASE11 [label="else"]; {rank=same "case 11" CASE11}; ;;DOT CASE11 -> "rmt:send-receive" [label="call failed"]; ;;DOT CASE11 -> "RESULT" [label="call succeeded"]; ;; not on homehost, do server query (else (extras-case-11 *default-log-port* runremote cmd params attemptnum rid))))) ;;DOT } ;; bunch of small functions factored out of send-receive to make debug easier ;; (define (extras-case-11 *default-log-port* runremote cmd params attemptnum rid) ;; (mutex-unlock! *rmt-mutex*) (debug:print-info 12 *default-log-port* "rmt:send-receive, case 9") ;; (mutex-lock! *rmt-mutex*) (let* ((conninfo (remote-conndat runremote)) (dat-in (condition-case ;; handling here has ;; caused a lot of ;; problems. However it ;; is needed to deal with ;; attemtped ;; communication to ;; servers that have gone ;; away (http-transport:client-api-send-receive 0 runremote cmd params) ;; (http-transport:client-api-send-receive 0 conninfo cmd params runremote) ((servermismatch) (vector #f "Server id mismatch" )) ((commfail)(vector #f "communications fail")) ((exn)(vector #f "other fail" (print-call-chain))))) (dat (if (and (vector? dat-in) ;; ... check it is a correct size (> (vector-length dat-in) 1)) dat-in (vector #f (conc "communications fail (type 2), dat-in=" dat-in)))) (success (if (vector? dat) (vector-ref dat 0) #f)) (res (if (vector? dat) (vector-ref dat 1) #f))) (if (and (vector? conninfo) (< 5 (vector-length conninfo))) (remote-last-access-set! runremote (current-seconds)) ;; refresh access time (begin (debug:print 0 *default-log-port* "INFO: Should not get here! conninfo=" conninfo) (set! conninfo #f) (http-transport:close-connections runremote))) (debug:print-info 13 *default-log-port* "rmt:send-receive, case 9. conninfo=" conninfo " dat=" dat " runremote = " runremote) (mutex-unlock! *rmt-mutex*) (if success ;; success only tells us that the transport was ;; successful, have to examine the data to see if ;; there was a detected issue at the other end (extras-transport-succeded *default-log-port* *rmt-mutex* attemptnum runremote res params rid cmd) (begin |
︙ | ︙ | |||
430 431 432 433 434 435 436 | (if qry-is-write (let ((start-time (current-seconds))) (mutex-lock! *db-multi-sync-mutex*) / (set! *db-last-access* start-time) ;; THIS IS PROBABLY USELESS? (we are on a client) (mutex-unlock! *db-multi-sync-mutex*))))) res)) | | < < < < < | | 404 405 406 407 408 409 410 411 412 413 414 415 416 417 418 419 420 | (if qry-is-write (let ((start-time (current-seconds))) (mutex-lock! *db-multi-sync-mutex*) / (set! *db-last-access* start-time) ;; THIS IS PROBABLY USELESS? (we are on a client) (mutex-unlock! *db-multi-sync-mutex*))))) res)) (define (rmt:send-receive-no-auto-client-setup runremote cmd run-id params) (let* ((run-id (if run-id run-id 0)) (res (http-transport:client-api-send-receive run-id runremote cmd params))) (if (and res (vector-ref res 0)) (vector-ref res 1) ;;; YES!! THIS IS CORRECT!! CHANGE IT HERE, THEN CHANGE rmt:send-receive ALSO!!! #f))) ;;====================================================================== ;; ;; A C T U A L A P I C A L L S |
︙ | ︙ | |||
468 469 470 471 472 473 474 | (define (rmt:login run-id) (rmt:send-receive 'login run-id (list *toppath* megatest-version (client:get-signature)))) ;; This login does no retries under the hood - it acts a bit like a ping. ;; Deprecated for nmsg-transport. ;; | | < | < < | 437 438 439 440 441 442 443 444 445 446 447 448 449 450 451 452 | (define (rmt:login run-id) (rmt:send-receive 'login run-id (list *toppath* megatest-version (client:get-signature)))) ;; This login does no retries under the hood - it acts a bit like a ping. ;; Deprecated for nmsg-transport. ;; (define (rmt:login-no-auto-client-setup runremote) (rmt:send-receive-no-auto-client-setup runremote 'login 0 (list *toppath* megatest-version (client:get-signature)))) ;; hand off a call to one of the db:queries statements ;; added run-id to make looking up the correct db possible ;; (define (rmt:general-call stmtname run-id . params) (rmt:send-receive 'general-call run-id (append (list stmtname run-id) params))) |
︙ | ︙ | |||
1065 1066 1067 1068 1069 1070 1071 | (debug:print-info 12 log-port "rmt:send-receive, case 3") (debug:print 0 log-port "WARNING: write transaction requested on a readonly area. cmd="cmd" params="params) #f) (define (extras-transport-failed *default-log-port* *rmt-mutex* attemptnum runremote cmd rid params) (debug:print 0 *default-log-port* "WARNING: communication failed. Trying again, try num: " attemptnum) (mutex-lock! *rmt-mutex*) | < | | 1031 1032 1033 1034 1035 1036 1037 1038 1039 1040 1041 1042 1043 1044 1045 | (debug:print-info 12 log-port "rmt:send-receive, case 3") (debug:print 0 log-port "WARNING: write transaction requested on a readonly area. cmd="cmd" params="params) #f) (define (extras-transport-failed *default-log-port* *rmt-mutex* attemptnum runremote cmd rid params) (debug:print 0 *default-log-port* "WARNING: communication failed. Trying again, try num: " attemptnum) (mutex-lock! *rmt-mutex*) (http-transport:close-connections runremote) (remote-server-url-set! runremote #f) (mutex-unlock! *rmt-mutex*) (debug:print-info 12 *default-log-port* "rmt:send-receive, case 9.1") (rmt:send-receive cmd rid params attemptnum: (+ attemptnum 1))) (define (extras-transport-succeded *default-log-port* *rmt-mutex* attemptnum runremote res params rid cmd) (if (and (vector? res) |
︙ | ︙ | |||
1094 1095 1096 1097 1098 1099 1100 | ;; server is ;; overloaded and we ;; want to ease off ;; the queries (let ((wait-delay (+ attemptnum (* attemptnum 10)))) (debug:print 0 *default-log-port* "WARNING: server is overloaded. Delaying " wait-delay " seconds and trying call again.") (mutex-lock! *rmt-mutex*) | | | 1059 1060 1061 1062 1063 1064 1065 1066 1067 1068 1069 1070 1071 1072 1073 1074 1075 1076 1077 | ;; server is ;; overloaded and we ;; want to ease off ;; the queries (let ((wait-delay (+ attemptnum (* attemptnum 10)))) (debug:print 0 *default-log-port* "WARNING: server is overloaded. Delaying " wait-delay " seconds and trying call again.") (mutex-lock! *rmt-mutex*) (http-transport:close-connections runremote) (set! *runremote* #f) ;; force starting over (mutex-unlock! *rmt-mutex*) (thread-sleep! wait-delay) (rmt:send-receive cmd rid params attemptnum: (+ attemptnum 1))) res)) ;; All good, return res #;(set-functions rmt:send-receive remote-server-url-set! http-transport:close-connections remote-conndat-set! debug:print debug:print-info remote-ro-mode remote-ro-mode-set! remote-ro-mode-checked-set! remote-ro-mode-checked) |