diff options
Diffstat (limited to 'rubbos/app/tomcat-connectors-1.2.32-src/native/TODO.txt')
| -rw-r--r-- | rubbos/app/tomcat-connectors-1.2.32-src/native/TODO.txt | 371 |
1 files changed, 0 insertions, 371 deletions
diff --git a/rubbos/app/tomcat-connectors-1.2.32-src/native/TODO.txt b/rubbos/app/tomcat-connectors-1.2.32-src/native/TODO.txt deleted file mode 100644 index 73dd13f3..00000000 --- a/rubbos/app/tomcat-connectors-1.2.32-src/native/TODO.txt +++ /dev/null @@ -1,371 +0,0 @@ - Licensed to the Apache Software Foundation (ASF) under one or more - contributor license agreements. See the NOTICE file distributed with - this work for additional information regarding copyright ownership. - The ASF licenses this file to You under the Apache License, Version 2.0 - (the "License"); you may not use this file except in compliance with - the License. You may obtain a copy of the License at - - http://www.apache.org/licenses/LICENSE-2.0 - - Unless required by applicable law or agreed to in writing, software - distributed under the License is distributed on an "AS IS" BASIS, - WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. - See the License for the specific language governing permissions and - limitations under the License. - -TODO for tomcat-connectors - -$Id: TODO.txt 920112 2010-03-07 20:57:35Z timw $ - -1) Optimize "distance" -====================== - -Sorting the list of balanced workers by distance would be nice, but: -How to combine the sorting with the offset implementation (especially -useful for strategy BUSYNESS under low load). - -Local error states and likely other features will also break in case -we do naive reordering. - -2) Reduce number of string comparisons in most_suitable -======================================================== - -a) redirect/domains - -It would be easy to improve the redirect string b an integer, giving the -index of the worker in the lb. Then lb would not need to search for the redirect worker. - -The same way, one could add a list with indizes to workers in the same domain. -Whenever domain names are managed (init and status worker update) one would -scan the worker list and update the index list. - -Finally one could have a list of workers, whose domain is the same as the redirect -attribute of the worker, because that's also something we consider. - -What I'm not sure about, even in the existing code, is the locking between updates -by the status worker and the process local information about the workers, -especially in the case, when status updates a redirect or domain attribute. - -I would like to keep these attributes and the new index arrays process local, -and the processes should find out about changes made by status to shm (redirect/domain) -and then rebuild their data. No need to get these on every request from the shm, -only the check for up-to-date should be made. - -b) exact matches for jvmRoutes - -Could we use hashes instead of string comparisons all the time? -I'm not sure, if a good enough hash takes longer than a string comparison though. - -3) Code separation between factory, validate and init in lb -============================================================ - -The factory contains: - - private_data->worker.retries = JK_RETRIES; - private_data->s->recover_wait_time = WAIT_BEFORE_RECOVER; - -I think, this should move to validate() or init(). -It might even be obsolete, because in init, we already have: - - pThis->retries = jk_get_worker_retries(props, p->s->name, - p->s->retries = pThis->retries; - p->s->recover_wait_time = jk_get_worker_recover_timeout(props, p->s->name, WAIT_BEFORE_RECOVER); - if (p->s->recover_wait_time < WAIT_BEFORE_RECOVER) - p->s->recover_wait_time = WAIT_BEFORE_RECOVER; - -Then: In validate there is - - p->lb_workers[i].s->error_time = 0; - -So shouldn't there also be - - p->lb_workers[i].s->maintain_time = time(NULL); - -4) Logging -========== - -a) Allow logging of request url or uuid in jk log to ease matching with access log. - -b) Implement log rotation for IIS. (done in 1.2.31) - -c) Allow adding of log notes for IIS like we do with Apache. - -d) Add error type info to access log notes - -e) Refactor: Use the same code files for the request logging functions in Apache 1.3 and 2.0. - -f) Refactor: Use the same code files for piped logging in Apache 1.3 and 2.0. - -5) ajpget -========== - -Combine ajplib and Apache ab to an ajp13 commandline client ajpget. - -6) Parsing workers.properties -============================= - -Parsing of workers.properties aditionally to just looking up attributes -would help users to detect syntax errors in the file. At the moment -no information will be logged, e.g. when attributes contain typos. - -Example: worker.list vs. worker.lists. - -7) Persisting workers.properties -================================ - -Make workers.properties persist from inside status worker. - -Add additional properties file, that contains a journal of property changes done -via the status worker. When initializing those overwrite the initial workers.properties. - -Update actions in the status worker will allow to optionally add a change to this journal. -We can also add a comment with timestamp etc. to each journal line. - -8) Reduce number of uses of time(NULL) -====================================== - -We use time(NULL) a lot. Since it only has resolution of a second, -I'm asking myself, if we could update the actual time in only a few -places and get time out of some variables when needed. The same does -not hold true for millisecond time, but in several cases we use the time, -it's not very critical, that it is exact. These cases are related to: - -Some of this is already been done, the remaining parts are: - -- last_access for usage against timeout value that is ~minutes -- error_time for usage against retry timeout that is ~minutes -- uri_worker_map checked for usage against JK_URIMAP_RELOAD=1 minute - -So I think, it would suffice to set an actual time at the beginning of -the request/response cycle (used by everything before the request is being -sent over the socket) and maybe after the response shows up/ an error occurs -(for everything else, if there is). - -For which cases would it be OK, to use the time before sending to TC: -- uri_worker_map "checked" (uri map lookup starts early) -- setting/testing last_access in - - jk_ajp_common.c:ajp_connect_to_endpoint() - - jk_ajp_common.c:ajp_get_endpoint() - - jk_ajp_common.c:ajp_maintain() - -What about the others: -- setting last_access in init should use the actual time in - jk_ajp_common.c:ajp_create_endpoint_cache() - -- setting last_access again after the service could also use the - actual time in jk_ajp_common.c:ajp_done() -- setting error_time should better use the actual time - jk_lb_worker.c service(): rec->s->error_time = time(NULL); - -The last two cases could again use the same time, which then would be needed -to be generated at the end or directly after service. - -9) Access/Modification Time in shm -================================== - -a) [Discussion] What will this generally be used for? At the moment, -only jk_status "uses" it, but it only sets the values, it never asks for them. - -b) [Improvement, minor] jk_shm_set_workers_time() implicitly calls -jk_shm_sync_access_time(), but jk_status does: - - jk_shm_set_workers_time(time(NULL)); - /* Since we updated the config no need to reload - * on the next request - */ - jk_shm_sync_access_time(); - -two times. So depending on the idea of the functionality of these calls, -either set_workers_time and sync_access_time should be independently, -or the second call in jk_status coulkd be removed. - -10) "Destroy" functionality -=========================== - -[Hint] Destroy on a worker never seems to free shm, -but I think that was already a flaw without shm. - -11) Locks against shm -===================== - -It might be an interesting experiment to implement an improved locking structure. -It looks like one would need in fact different types of locks. -In shm we have as read/write information: - -Changed only by status worker: -- redirect, domain, lb_factor, sticky_session, sticky_session_force, - recover_wait_time, retries, status (is_disabled, is_stopped). - -These changes need some kind of reconfiguration in the threads after -change and before the next request/response. Since changes are rare, -here we would be better of, with a simple detect change and copy from -shm to process procedure. status updates the data in shm and after that -the time stamp on the shh. Each process checks the time stamp before -doing a request, and when the time stamp changed it does a writer CS -lock and updates it's local copy. All threads always do a reader CS -lock when doing a request/response cycle. Reader CS locks are concurrent, -writers are exclusive. So readers are not allowed, when the config data is being updated. - -Changed by the threads themselves (and via reset by the status worker): -- counters needed by routing decisions (lb_value, readed, transferred, busy) -- timers needed by maintenance functions (error_time, servic_time/maintain_time) -- status is_busy, in_error_state -- uncritical data with no influence on routing decisions (max_busy, elected, errors, - in_recovering) - -Here again we could improve by using reader/writer locks. I have a -tendency for the PESSIMISTIC side of locking, but I think we could -shrink the code blocks that should be locked. At the monent they are -pretty big (most of get_most_suitable_worker). - -Read-only: name and id. - -By the way: at several places we don't check for errors on getting the lock. - -12) Global locks -================ - -We might want to make the lock technology choosable, like httpd does. -E.g. on Solaris the default lock type if fcntl, and we can easily -get an invalid EDEADLOCK for our jk_log_lock. - -The following pthread based non global locks are used: - -- one mutex for each AJP worker, synchronizing access to the connection -pool, which exists per process - -- one mutex for each lb worker - -- a mutex used during dynamic update of uriworkermap.properties to -prevent concurrent updates. Updates are done per process. - -- a mutex to prevent concurrent execution of the process local internal -maintenance task - -- a mutex for access to the shared memory when changing or reading -configuration parameters. That might be a little unsafe, because it -actually should be a global mutex, not a process local, but those config -changes are only done due to interaction with the status worker, so -there's very little chance for unwanted concurrency here. All dynamic -runtime data are already marked as being volatile. - -All except the last seem to be safe. The last might need some hybrid model -using thread local mostly and process global when doing updates. - -See also: http://marc.info/?t=123394059800003&r=1&w=2 - -13) Understand the exact behaviour of shm and restarts -====================================================== - -Furthermore: rotatelogs (?) and gzip (mod_mime_magic) seem to close -the (non-existing) shm file. Maybe a problem on some platforms? - -14) What I didn't yet check -=========================== - -a) Correctness of is_busy handling - -b) Correctness of the reset values after reset by status worker - -c) What would be the exact behaviour, if shm does not work (memory case). - Will this be a critical failure, or will we only experience a - degradation in routing decisions. - -d) How complete is mod_proxy_ajp/mod_proxy_balancer. - Port changes from mod_jk to them. - -15) Status worker -================= - -Allow managing pool and connection parameters. Add flags to -pool and connections to signal workers and maintenance whether -existing connections should be closed and renewed. - -Check completeness of attribute manageability for AJP workers. - -Check completeness of runtime data display, e.g. -reset_time, recover_time, etc. Maybe also add "last error type". - -Work on a global display of process local data, e.g. state of the -process local connection pools (sizes, num connected/idle). - -Rework the GUI: - -- Basic overview start page with links to the workers, - maybe a checkbox to decide whether you want to see config data too. -- Detail view (what type of error was last and when etc.), detail view - for connection pools. - -16) URI mapping -=============== - -Add more extensions? - -17) Connection Pool -=================== - -How would a good global maximum count look like? -Simply limit busyness? Soft limit (applies only to non-sticky requests) -and a hard limit (applies to all requests)? - -18) IP V6 -========= - -There's a Bugzilla with a patch. - -19) IIS Chunked Encoding -======================== - -Move from alternative build to default. - -What about the other ifdef'd features? - -20) Add REMORE_PORT as a default JkEnvVar -========================================= - -... and port to IIS to fix the getRemotePort() problem. - -21) Rework HowTo docs -===================== - -22) Add better example config -============================= - -23) Remove JNI worker -===================== - -24) Watchdog reload of uriworkermap.properties -============================================== - -Questions about uriworkermap.properties watchdog reload (r745898): -- shm lock needed? -- Apache port (need to iterate over vhosts) - -25) Watchdog backend probing -============================ - -Configurable probe URL to test backend, e.g. to decide about -recovery instead of using random real requests. - -26) Commandline shm -=================== - -Commandline tool to read data from shm file. - -27) Status worker property format -================================= - -Check whether we return list properties as one line -(because Java doesn't allow the same property key multiple -times). - -Applies possibly to: - -"list", -BALANCE_WORKERS, -MOUNT_OF_WORKER, -USER_OF_WORKER, -GOOD_RATING_OF_WORKER, -BAD_RATING_OF_WORKER, -STATUS_FAIL_OF_WORKER, - |