Commit Graph

178 Commits

Author SHA1 Message Date
Christian Schwarz
1a8d2c5ebe replication: context support and propert closing of stale readers 2018-07-08 23:31:46 +02:00
Christian Schwarz
8cca0a8547 Initial working version
Summary:
* Logging is still bad
* test output in a lot of placed
* FIXMEs every where

Test Plan: None, just review

Differential Revision: https://phabricator.cschwarz.com/D2
2018-06-24 10:44:00 +02:00
Christian Schwarz
0918ef6815 WIP: diffing and replication algorithm 2018-05-02 21:26:24 +02:00
Christian Schwarz
9d7110eaad config: fix shadowed error return values 2018-04-14 11:25:12 +02:00
Christian Schwarz
82ea535692 daemon: expose prometheus in new global.monitoring config section + document it
refs #67
2018-04-14 11:24:47 +02:00
Christian Schwarz
a4da029105 cmd: prometheus job type and Task instrumentation
refs #67
2018-04-13 23:37:53 +02:00
Christian Schwarz
aa3865d0a3 daemon: Job types as dedicated type
refs #67
2018-04-05 22:22:55 +02:00
Christian Schwarz
0895e02844 daemon: Task: track relation to parent job
refs #67
2018-04-05 22:18:22 +02:00
Christian Schwarz
26b436463d ssh+stdinserver: connect: dial_timeout
This  is a follow-up to ccd062e
2018-03-04 17:19:41 +01:00
Christian Schwarz
aa92261ea7 bookmarking: prune policy for bookmarks
refs #34
2018-02-17 20:48:31 +01:00
Christian Schwarz
8e34843eb1 autosnap: do not treat zero fs filter results as fatal 2018-02-17 19:27:00 +01:00
Christian Schwarz
bfaf6fdfbb daemon: fix missing newline on parse error 2018-02-17 17:43:55 +01:00
Christian Schwarz
f992fed968 control pprof rewrite: expose pprof metrics via HTTP server controlled from CLI 2018-02-17 16:20:10 +01:00
Christian Schwarz
94967b596c docs: document changes to ssh+stdinserver transport implementation: ccd062e 2018-02-17 15:16:29 +01:00
Christian Schwarz
f3d3a7f5f8 stdinserver: fixup ccd062e: assert socket is in private directory 2018-02-17 14:12:44 +01:00
Christian Schwarz
ccd062e238 ssh+stdinserver: dump sshbytestream for github.com/problame/go-netssh
Cleaner abstractions + underlying go-rwccmd package does proper handling
of asynchronous exits, etc.
2018-02-17 01:08:15 +01:00
Christian Schwarz
6b5bd0a43c job pull + source: fix broken connection teardown
Issue #56 shows zombie SSH processes.
We fix this by actually Close()ing the RWC in job pull.
If this fixes #56 it also fixes #6 --- it's the same issue.

Additionally, debugging around this revealed another issue: just
Close()ing the sshbytestream in job source will apparently outpace the
normal data stream of stdin and stdout (URG or PUSH flags?).  leading
to ugly errors in the logs.
With proper TCP connections, we would simply set the connection to
linger and close it, letting the kernel handle the final timeout. Meh.

refs #56
refs #6
2018-02-16 20:57:27 +01:00
Christian Schwarz
921bccb960 job source: use task logger 2018-02-15 23:51:57 +01:00
Christian Schwarz
5f2c14adab zfs: use custom datatype to pass ZFS properties in ZFSSet
refs #55
2018-01-05 18:42:10 +01:00
Christian Schwarz
787675aee8 control status command: only show verbose logs on user request 2017-12-30 13:53:19 +01:00
Christian Schwarz
01e0519b7b control status subcommand: fix typo in usage 2017-12-30 13:44:55 +01:00
Christian Schwarz
8742b7f763 handler: fix typo in log message 2017-12-30 13:29:04 +01:00
Christian Schwarz
56f13741f9 test pattern subcommand: better example command 2017-12-29 22:45:38 +01:00
Christian Schwarz
61842988b9 Task & TaskStatus: DeepCopy(): actually copy lastUpdate field
otherwise, only changes to activity level would udpate TaskStatus
LastUpdate field

refs #10
2017-12-29 21:43:12 +01:00
Christian Schwarz
be7176bee7 Puller: fix wrong filesystem log field usage
was introduced in 9465b593
2017-12-29 21:25:42 +01:00
Christian Schwarz
839eccf513 logger.Outlet: WriteEntry must not block
- make TCPOutlet fully asynchronous, dropping messages if connection is
  not fast enough
- syslog is just fine for now, local anyways
- stdout same thing

refs #26
2017-12-29 17:21:58 +01:00
Christian Schwarz
acd9aedb98 cmd control status: unify job logs, option to show only one job & always show logs
refs #10
2017-12-27 18:34:24 +01:00
Christian Schwarz
835cf6b12f cmd control status: warn about inactive tasks
refs #10
2017-12-27 18:34:24 +01:00
Christian Schwarz
4b3d83ec1f TaskStatus: add LastUpdate field
refs #10
2017-12-27 18:34:24 +01:00
Christian Schwarz
d13c6e3fc3 job local: refactor + use Task API
refs #10
2017-12-27 18:34:24 +01:00
Christian Schwarz
63fa7a67e9 job source: refactor + use Task API
refs #10
2017-12-27 18:34:24 +01:00
Christian Schwarz
7d89d1fb00 job pull: refactor + use Task API
refs #10
2017-12-27 18:34:24 +01:00
Christian Schwarz
b69089a527 Puller: refactor + use Task API
* drop rx byte count functionality
* will be re-added to Task as necessary

refs #10
2017-12-27 14:39:47 +01:00
Christian Schwarz
59e34942d1 Puller: make main interface public
refs #10
2017-12-27 14:39:46 +01:00
Christian Schwarz
91c4a97f72 Pruner: refactor + use Task API
refs #10
2017-12-27 14:39:46 +01:00
Christian Schwarz
13562b48ed IntervalAutosnap: refactor + use Task API
refs #10
2017-12-27 14:39:46 +01:00
Christian Schwarz
58ee796394 adopt Task API: infect datastructures
refs #10
2017-12-27 14:39:46 +01:00
Christian Schwarz
ce351146cf job control: implement JobStatus 2017-12-27 14:39:46 +01:00
Christian Schwarz
14b8d69a63 cmd control status + expose DaemonStatus via control API
refs #10
2017-12-27 14:39:46 +01:00
Christian Schwarz
8c7e373049 daemon: DaemonStatus + JobStatus + dummy implementation
refs #10
2017-12-27 14:39:46 +01:00
Christian Schwarz
2c87b15e83 daemon: Task abstraction + TaskStatus
An instance of Task tracks a single thread of activity that is part of a Job.

While the docs already use this terminology of tasks being composed of jobs,
the code did not have an object to represent these semantics.
Now it does:

* A task t is initialized with a root activity, which is its name
* t can t.Enter() and t.Finish() an activity, building
  a stack of activities
* t's code can get a logger t.Log() whose logTaskField is set to the
  concatenated stack of activities
* t's code can update IO progress it made since leaving idle state
* t's code's log output vie t.Log() is captured since leaving idle
  state
  * FIXME: find a way to bound that buffer

refs #10
refs #48
2017-12-27 14:39:46 +01:00
Christian Schwarz
d7f3fb93ae bash completions: hidden subcommand + integrate into Makefile 2017-12-27 14:39:46 +01:00
Christian Schwarz
ebf209427a logging: support ignoring fields in HumanFormatter
should be refactored to logger one day so the implementation of ignoring
is not duplicated to each outlet.

refs #10
2017-12-27 14:39:46 +01:00
Christian Schwarz
261d095108 logger: support forking of outlets
refs #10
2017-12-27 13:50:07 +01:00
Christian Schwarz
583a63a68f refactor: encapsulate pulling in a struct
refs #10
2017-12-24 15:23:28 +01:00
Christian Schwarz
896f31bbf3 'zrepl version' and 'zrepl control version' subcommand + maintainer README
Version is autodetected on build using git
If it cannot be detected with git, an override must be provided.

For tracability of distros, the distroy packagers should override as
well, which is why I added a README entry for package mainatiners.

refs #35
2017-11-18 21:12:48 +01:00
Christian Schwarz
bfbab9382e fixup: remove unused StdoutOutlet function
refs #28
2017-11-17 00:36:48 +01:00
Christian Schwarz
2bfcfa5be8 logging: first outlet receives logger error message
Abandons stderr special-casing:

* looks weird on shell and IO redirection to same file because of
interleaving of stdout and stderr
* better than a separate dedicated outlet because it does not require
additional configuration

fixes #28

BREAK SEMANTICS CONFIG
2017-11-17 00:25:38 +01:00
Christian Schwarz
a7f70a566d logger: write internal / outlet errors to an error outlet
refs #28
2017-11-16 23:49:47 +01:00
Christian Schwarz
b576253ea8 logging: fixup 4763486: implementation would parse 'date' instead of 'time' field in config 2017-11-15 11:14:20 +01:00