prevents out of memory situation if many IP's makes extremely many failures (or very large files since last fail2ban run);
closes gh-1277
todo: parameter `maxentries` should be configurable (jail.conf resp. fail2ban.conf);
todo: adjust ban-time-incr branch by merge (table "bips").
```
in executeCmd
if retcode < 0:
TypeError: unorderable types: NoneType() < int()
```
2) prevents a rarely test case bug of testExecuteTimeoutWithNastyChildren, because no stdout (Resource temporarily unavailable), possible no flush by IO of the killing process;
definitely closes gh-161, also other usage of asyncore event loop (in test_smtp.py)
repair cache in ipToName (can returns None), precaching of invalid IPs (according to RFC 5737) to stop endless wait for resolving it in test cases.
- new options for "fail2ban-testcases" introduced: "-g" or "--no-gamin" and "-m" or "--memory-db", both are true also if "-f" or "--fast" specified,
for example: `fail2ban-testcases -ngm` will runs faster (because no network, no gamin, memory database)
but will use the same default sleep intervals as in production (in comparison to -nf);
- seekToTime rewritten, accuracy increased by seekToTime, extended for all FileFilter (PyInotify, Gamin also), test cases extended etc.
- common performance optimized and code reviewed;
-- cache dnsToIp, ipToName to prevent long wait during retrieving of ip/name for wrong dns or lazy dns-system;
-- instead of simple "sleep" used conditional wait "wait_for", that internal increases sleep interval up to sleeptime;
-- ticket / banmanager / failmanager modules are performance optimized;
-- api of filter (log files), jail, etc. rewritten and extended for performance purposes;
- performance of test cases optimized:
-- added option "--fast" to decrease wait intervals, avoid passive waiting, and skip few very slow test cases;
- code review after partially cherry pick of branch 'ban-time-incr' (see gh-716)
-- ticket module prepared to easy merge with newest version of 'ban-time-incr', now additionally holds banTime, banCount and json-data;
-- executeCmd partially moved from action to new module utils, etc.
-- python 2.6 compatibility;
- testExecuteTimeoutWithNastyChildren: test case repaired - wait for pid file inside bash, kill tree in any case (gh-1155);
- testSocket: test case repaired - wait for server thread starts a socket (listener)
addJail (executed before addLog) early uses a "INSERT OR REPLACE" statement to update "enabled" to 1 (and add jail the first time used at once), but this syntax in sqlite always deletes an entry (cause of constraint) and inserts it again, so because of CASCADE all log entries with this jail will be also deleted from logs table.
These fixes are pretty pedantic, but they do simplify the script a
little.
* Checking the existence of a file/directory before creating/deleting
it adds complexity and raciness. There are better options.
* mkdir -p does the job of making sure a directory exists. (It only
fails if there's a filesystem error or something.)
* Likewise, rm -f doesn't fail if the file doesn't exist.
* rm -r isn't neccessary because the socket shouldn't be a directory.
(If it is for some reason, that should be an error.)
sebres