Overview
Comment: | Moved init of runremote to where (make-remote) was called. On discarding connections ensure they are closed. Added annotation for study/debug of the flow through rmt:send-receive. Run make rmt.pdf to create it. |
---|---|
Downloads: | Tarball | ZIP archive | SQL archive |
Timelines: | family | ancestors | descendants | both | v1.64 |
Files: | files | file ages | folders |
SHA1: |
d46d2249c449d390c75a5fbfe4694c2c |
User & Date: | matt on 2017-05-29 22:28:36 |
Other Links: | branch diff | manifest | tags |
Context
2017-05-30
| ||
01:12 | Added support for triggers in megatest.config. Updated docs with new variables on the command line for trigger scripts. Triggers are now run via nbfake. check-in: ad27863571 user: matt tags: v1.64 | |
2017-05-29
| ||
22:28 | Moved init of runremote to where (make-remote) was called. On discarding connections ensure they are closed. Added annotation for study/debug of the flow through rmt:send-receive. Run make rmt.pdf to create it. check-in: d46d2249c4 user: matt tags: v1.64 | |
2017-05-28
| ||
00:49 | added -testdat-csv option to megatest; fixed -set-values so that tolerance and expected value are not required fields; added delay in set-values to allow for delayed write on server start check-in: a5adfa3f9a user: bjbarcla tags: v1.64 | |
Changes
Modified Makefile from [024b5aa496] to [ce2a9bd188].
︙ | ︙ | |||
302 303 304 305 306 307 308 | if csi -ne '(use postgresql)';then \ echo "(use postgresql)(hash-table-set! *available-db* 'postgresql #t)" >> altdb.scm;\ fi portlogger-example : portlogger-example.scm api.o archive.o client.o common.o configf.o daemon.o dashboard-tests.o db.o dcommon.o ezsteps.o filedb.o genexample.o gutils.o http-transport.o items.o keys.o launch.o lock-queue.o margs.o megatest-version.o mt.o ods.o portlogger.o process.o rmt.o rpc-transport.o runconfig.o runs.o sdb.o server.o synchash.o tasks.o tdb.o tests.o tree.o csc $(CSCOPTS) portlogger-example.scm api.o archive.o client.o common.o configf.o daemon.o dashboard-tests.o db.o dcommon.o ezsteps.o filedb.o genexample.o gutils.o http-transport.o items.o keys.o launch.o lock-queue.o margs.o megatest-version.o mt.o ods.o portlogger.o process.o rmt.o rpc-transport.o runconfig.o runs.o sdb.o server.o synchash.o tasks.o tdb.o tests.o tree.o | > > > > | 302 303 304 305 306 307 308 309 310 311 312 | if csi -ne '(use postgresql)';then \ echo "(use postgresql)(hash-table-set! *available-db* 'postgresql #t)" >> altdb.scm;\ fi portlogger-example : portlogger-example.scm api.o archive.o client.o common.o configf.o daemon.o dashboard-tests.o db.o dcommon.o ezsteps.o filedb.o genexample.o gutils.o http-transport.o items.o keys.o launch.o lock-queue.o margs.o megatest-version.o mt.o ods.o portlogger.o process.o rmt.o rpc-transport.o runconfig.o runs.o sdb.o server.o synchash.o tasks.o tdb.o tests.o tree.o csc $(CSCOPTS) portlogger-example.scm api.o archive.o client.o common.o configf.o daemon.o dashboard-tests.o db.o dcommon.o ezsteps.o filedb.o genexample.o gutils.o http-transport.o items.o keys.o launch.o lock-queue.o margs.o megatest-version.o mt.o ods.o portlogger.o process.o rmt.o rpc-transport.o runconfig.o runs.o sdb.o server.o synchash.o tasks.o tdb.o tests.o tree.o # create a pdf dot graphviz diagram from notations in rmt.scm rmt.pdf : rmt.scm grep ';;DOT' rmt.scm | sed -e 's/.*;;DOT //' > rmt.dot;dot -Tpdf rmt.dot -o rmt.pdf |
Modified rmt.scm from [f26cf937f2] to [f39fa419cf].
︙ | ︙ | |||
46 47 48 49 50 51 52 53 54 | (define *send-receive-mutex* (make-mutex)) ;; should have separate mutex per run-id ;; RA => e.g. usage (rmt:send-receive 'get-var #f (list varname)) ;; (define (rmt:send-receive cmd rid params #!key (attemptnum 1)(area-dat #f)) ;; start attemptnum at 1 so the modulo below works as expected ;; do all the prep locked under the rmt-mutex (mutex-lock! *rmt-mutex*) | > > > > > > | > > > | | | | | > > > > | | | | | | > > > > > | | > > > | | | | > > > > > > > > > > > > > | | | | > > > > | > > > | | > > > | | | | | | > > > | < > > > > | > > > > | 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60 61 62 63 64 65 66 67 68 69 70 71 72 73 74 75 76 77 78 79 80 81 82 83 84 85 86 87 88 89 90 91 92 93 94 95 96 97 98 99 100 101 102 103 104 105 106 107 108 109 110 111 112 113 114 115 116 117 118 119 120 121 122 123 124 125 126 127 128 129 130 131 132 133 134 135 136 137 138 139 140 141 142 143 144 145 146 147 148 149 150 151 152 153 154 155 156 157 158 159 160 161 162 163 164 165 166 167 168 169 170 171 172 173 174 175 176 177 178 179 180 181 182 183 184 185 186 187 188 189 190 191 192 193 194 195 196 197 198 199 200 201 202 203 204 205 206 207 208 209 210 211 212 213 214 215 216 217 218 219 220 221 222 223 224 225 226 227 228 229 230 231 232 233 234 235 236 237 238 239 240 | (define *send-receive-mutex* (make-mutex)) ;; should have separate mutex per run-id ;; RA => e.g. usage (rmt:send-receive 'get-var #f (list varname)) ;; (define (rmt:send-receive cmd rid params #!key (attemptnum 1)(area-dat #f)) ;; start attemptnum at 1 so the modulo below works as expected ;;DOT digraph megatest_state_status { ;;DOT ranksep=0; ;;DOT // rankdir=LR; ;;DOT node [shape="box"]; ;;DOT "rmt:send-receive" -> MUTEXLOCK; ;;DOT { edge [style=invis];"case 1" -> "case 2" -> "case 3" -> "case 4" -> "case 5" -> "case 6" -> "case 7" -> "case 8" -> "case 9" -> "case 10" -> "case 11"; } ;; do all the prep locked under the rmt-mutex (mutex-lock! *rmt-mutex*) ;; 1. check if server is started IFF cmd is a write OR if we are not on the homehost, store in runremote ;; 2. check the age of the connections. refresh the connection if it is older than timeout-20 seconds. ;; 3. do the query, if on homehost use local access ;; (let* ((start-time (current-seconds)) ;; snapshot time so all use cases get same value (areapath *toppath*);; TODO - resolve from dbstruct to be compatible with multiple areas (runremote (or area-dat *runremote*)) (readonly-mode (if (and runremote (remote-ro-mode-checked runremote)) (remote-ro-mode runremote) (let* ((dbfile (conc *toppath* "/megatest.db")) (ro-mode (not (file-write-access? dbfile)))) ;; TODO: use dbstruct or runremote to figure this out in future (if runremote (begin (remote-ro-mode-set! runremote ro-mode) (remote-ro-mode-checked-set! runremote #t) ro-mode) ro-mode))))) ;; DOT INIT_RUNREMOTE; // leaving off - doesn't really add to the clarity ;; DOT MUTEXLOCK -> INIT_RUNREMOTE [label="no remote?"]; ;; DOT INIT_RUNREMOTE -> MUTEXLOCK; ;; ensure we have a record for our connection for given area (if (not runremote) ;; can remove this one. should never get here. (begin (set! *runremote* (make-remote)) (set! runremote *runremote*))) ;; new runremote will come from this on next iteration ;; DOT SET_HOMEHOST; // leaving off - doesn't really add to the clarity ;; DOT MUTEXLOCK -> SET_HOMEHOST [label="no homehost?"]; ;; DOT SET_HOMEHOST -> MUTEXLOCK; ;; ensure we have a homehost record (if (not (pair? (remote-hh-dat runremote))) ;; not on homehost (thread-sleep! 0.1) ;; since we shouldn't get here, delay a little (remote-hh-dat-set! runremote (common:get-homehost))) ;;(print "BB> readonly-mode is "readonly-mode" dbfile is "dbfile) (cond ;;DOT EXIT; ;;DOT MUTEXLOCK -> EXIT [label="> 15 attempts"]; {rank=same "case 1" "EXIT" } ;; give up if more than 15 attempts ((> attemptnum 15) (debug:print 0 *default-log-port* "ERROR: 15 tries to start/connect to server. Giving up.") (exit 1)) ;;DOT CASE2 [label="local\nreadonly\nquery"]; ;;DOT MUTEXLOCK -> CASE2; {rank=same "case 2" CASE2} ;;DOT CASE2 -> "rmt:open-qry-close-locally"; ;; readonly mode, read request- handle it - case 2 ((and readonly-mode (member cmd api:read-only-queries)) (mutex-unlock! *rmt-mutex*) (debug:print-info 12 *default-log-port* "rmt:send-receive, case 2") (rmt:open-qry-close-locally cmd 0 params) ) ;;DOT CASE3 [label="write in\nread-only mode"]; ;;DOT MUTEXLOCK -> CASE3 [label="readonly\nmode?"]; {rank=same "case 3" CASE3} ;;DOT CASE3 -> "#f"; ;; readonly mode, write request. Do nothing, return #f (readonly-mode (mutex-unlock! *rmt-mutex*) (debug:print-info 12 *default-log-port* "rmt:send-receive, case 3") (debug:print 0 *default-log-port* "WARNING: write transaction requested on a readonly area. cmd="cmd" params="params) #f) ;;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) (let ((expire-time (+ (- start-time (remote-server-timeout runremote))(random 10)))) ;; Subtract or add the random value? Seems like it should be substract but Neither fixes the "WARNING: failure in with-input-from-request to #<request>.\n message: Server closed connection before sending response" (< (http-transport:server-dat-get-last-access (remote-conndat runremote)) expire-time))) (debug:print-info 0 *default-log-port* "Connection to " (remote-server-url runremote) " expired due to no accesses, forcing new connection.") (let ((old-conn (remote-conndat runremote))) (handle-exceptions exn (begin (print-call-chain *default-log-port*) (debug:print-error 0 *default-log-port* " closing connection failed with error: " ((condition-property-accessor 'exn 'message) exn))) (http-transport:close-connections area-dat: runremote))) (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"; ;; on homehost and this is a read ((and (not (remote-force-server runremote)) ;; honor forced use of server, i.e. server NOT required (cdr (remote-hh-dat runremote)) ;; on homehost (member cmd api:read-only-queries)) ;; this is a read (mutex-unlock! *rmt-mutex*) (debug:print-info 12 *default-log-port* "rmt:send-receive, case 5") (rmt:open-qry-close-locally cmd 0 params)) ;;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 ((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)))) ;; server has died. NOTE: this is not a cheap call! Need better approach. (set! *runremote* (make-remote)) (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 (not (member cmd api:read-only-queries)) ;; this is a write (remote-server-url runremote)) ;; have a server (mutex-unlock! *rmt-mutex*) (debug:print-info 12 *default-log-port* "rmt:send-receive, case 4.1") (rmt:open-qry-close-locally cmd 0 params)) ;;DOT CASE8 [label="force\nserver"]; ;;DOT MUTEXLOCK -> CASE8 [label="server not required,\nhave homehost info,\nno connection yet,\nnot a read-only query"]; {rank=same "case 8" CASE8}; ;;DOT CASE8 -> "rmt:open-qry-close-locally"; ;; on homehost, no server contact made and this is a write, passively start a server ((and (not (remote-force-server runremote)) ;; honor forced use of server, i.e. server NOT required (cdr (remote-hh-dat runremote)) ;; have homehost (not (remote-server-url runremote)) ;; no connection yet (not (member cmd api:read-only-queries))) ;; not a read-only query (debug:print-info 12 *default-log-port* "rmt:send-receive, case 8") (let ((server-url (server:check-if-running *toppath*))) ;; (server:read-dotserver->url *toppath*))) ;; (server:check-if-running *toppath*))) ;; Do NOT want to run server:check-if-running - very expensive to do for every write call (if server-url (remote-server-url-set! runremote server-url) ;; the string can be consumed by the client setup if needed (if (common:force-server?) (server:start-and-wait *toppath*) (server:kind-run *toppath*)))) (remote-force-server-set! runremote (common:force-server?)) (mutex-unlock! *rmt-mutex*) (debug:print-info 12 *default-log-port* "rmt:send-receive, case 8.1") (rmt:open-qry-close-locally cmd 0 params)) ;;DOT CASE9 [label="force server\nnot on homehost"]; ;;DOT MUTEXLOCK -> CASE9 [label="no connection\nand either require server\nor not on homehost"]; {rank=same "case 9" CASE9}; ;;DOT CASE9 -> "start\nserver" -> "rmt:send-receive"; ((or (and (remote-force-server runremote) ;; we are forcing a server and don't yet have a connection to one (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*)) ;; 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 (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 (case (remote-transport runremote) |
︙ | ︙ | |||
214 215 216 217 218 219 220 221 222 223 224 225 226 227 | (remote-conndat-set! runremote #f) (remote-server-url-set! runremote #f) (debug:print-info 12 *default-log-port* "rmt:send-receive, case 9.1") (if (not (server:check-if-running *toppath*)) (server:start-and-wait *toppath*)) (rmt:send-receive cmd rid params attemptnum: (+ attemptnum 1))))))))) ;; (define (rmt:update-db-stats run-id rawcmd params duration) ;; (mutex-lock! *db-stats-mutex*) ;; (handle-exceptions ;; exn ;; (begin ;; (debug:print 0 *default-log-port* "WARNING: stats collection failed in update-db-stats") ;; (debug:print 0 *default-log-port* " message: " ((condition-property-accessor 'exn 'message) exn)) | > > | 268 269 270 271 272 273 274 275 276 277 278 279 280 281 282 283 | (remote-conndat-set! runremote #f) (remote-server-url-set! runremote #f) (debug:print-info 12 *default-log-port* "rmt:send-receive, case 9.1") (if (not (server:check-if-running *toppath*)) (server:start-and-wait *toppath*)) (rmt:send-receive cmd rid params attemptnum: (+ attemptnum 1))))))))) ;;DOT } ;; (define (rmt:update-db-stats run-id rawcmd params duration) ;; (mutex-lock! *db-stats-mutex*) ;; (handle-exceptions ;; exn ;; (begin ;; (debug:print 0 *default-log-port* "WARNING: stats collection failed in update-db-stats") ;; (debug:print 0 *default-log-port* " message: " ((condition-property-accessor 'exn 'message) exn)) |
︙ | ︙ |