fang
7a4034f48a
Merge pull request #5977 from urbit/m/cork-undone
...
ames, gall: no %done on corks
2022-09-06 22:17:34 +02:00
fang
9c931f844b
Merge pull request #5976 from urbit/m/ames-restir
...
ames: clean & re-set recork timer during %stir
2022-09-06 22:17:18 +02:00
fang
36dd2f8ace
Merge branch 'next/arvo' into m/recork-slower
2022-09-06 22:16:41 +02:00
fang
fec3c45d30
Merge pull request #5971 from urbit/jb/recork-error
...
ames: handle failed recork timers
2022-09-06 22:15:12 +02:00
fang
0e5b3b4107
ames: don't give %done on cork
...
Locally, %cork is guaranteed to succeed instantly. The requesting vane
doesn't need an ack to safely consider the flow "closed".
2022-08-25 18:13:29 +02:00
fang
81462da7aa
ames: clean & re-set recork timer during %stir
...
Pre-release ships may have multiple recork timers. This ensures that,
after stir, we always have exactly one.
2022-08-25 14:42:16 +02:00
fang
0871c24e24
ames: bump recork timer back up to a day
...
We had lowered this initially for easier testing, but a day should be
sufficient for real-world use.
2022-08-25 14:11:45 +02:00
fang
b2422030f7
ames: correctly identify stale routes
2022-08-25 13:41:55 +02:00
Joe Bryan
76cc8939d3
ames: fix error slog on failed attestation timer
2022-08-23 14:34:08 -04:00
Joe Bryan
5b8f26558e
ames: handle failed recork timers
2022-08-23 14:33:53 -04:00
Joe Bryan
679b38138d
ames: prevent larval stage from possibly clobbering adult state
2022-08-23 10:37:01 -04:00
Ted Blackman
5a96f3bdbe
ames: drop %hear as larva; check duct before metamorpohse
2022-08-23 16:45:17 +03:00
Ted Blackman
07601c6bee
ames: metamorphose on first +take if possible
2022-08-23 16:09:35 +03:00
Ted Blackman
9abe49489d
ames: +on-stir clean up comments
2022-08-23 13:31:10 +03:00
Joe Bryan
9439841ce8
ames: refactor %stir, %rest stale/wrong timers
2022-08-22 20:14:02 -04:00
fang
3a7938f057
ames: say "metamorphosis" in all relevant cases
2022-08-20 00:05:20 +02:00
fang
53b0e7ecaa
ames: ensure larva initializes adult's state
...
Previously, when the larva got to processing enqueued events, it was
doing so without loading state into the adult beforehand, resulting in
incorrect processing of events.
Here, we make the larva call +molt more eagerly, ensuring that the adult
always has its state available when we use it.
2022-08-19 23:58:10 +02:00
fang
ce2e433a8a
ames: don't skip closing flows in +on-stir
...
Yes, there is a global timer for closing flows, but all that does is
enqueue a cork message. +on-stir needs to set _pump_ timers for all
flows that might still have messages to send, which includes closing
flows.
2022-08-19 20:26:08 +02:00
Ted Blackman
6b836bb56d
ames: fix no-op for messages on closing bones
2022-08-10 20:18:57 +03:00
Ted Blackman
20ccac22a2
ames: clean up closing bone check and printfs
2022-08-10 19:55:07 +03:00
Ted Blackman
0baad53fb4
ames: drop pleas on dead bones
2022-08-10 19:47:59 +03:00
Ted Blackman
ce9f692ecd
ames: clean up %cork plea handling
...
+on-plea gets called in two very different ways:
1) handling request from local vane to send %plea to peer
2) handling %cork request from another ship, which our local ames has %pass'ed
to ourselves
In the second case, we shouldn't print misleadingly, or bind a duct in the ossuary.
2022-08-10 00:47:38 +03:00
Ted Blackman
2dc7171dfa
ames: no-op on acks on corked bones
2022-08-10 00:44:54 +03:00
Ted Blackman
da591a4078
ames: invert busted conditional in recork timer
2022-08-05 02:37:03 +03:00
Ted Blackman
56d4906a13
ames: try next recork on cork ack
...
Without this, a ship would send a cork on a max of one flow per
recork timer, which could take years to clear for some ships.
This starts a hot loop of trying the next cork once one gets
positively acked.
2022-08-05 01:35:34 +03:00
Ted Blackman
54cd1a5eca
ames: simplify recork timer
...
The previous recork timer queued up %cork messages without sending them.
It also relied on making sure pump timers didn't get set for recork bones.
This was fragile.
The new design enqueues up to one new %cork message per ship during each
recork timer, based on the state of the flow. If the flow is closing but
there are no outstanding messages in it, then it needs to be recorked.
Flows will be recorked in ascending numerical order by bone.
2022-08-05 01:13:21 +03:00
fang
9433d3354d
ames: unsent-fragments is a list, not a tree
...
Measure accordingly.
2022-08-03 21:03:49 +02:00
fang
7af527e605
ames: correctly measure queue size
2022-08-03 20:52:03 +02:00
fang
e6453bae3c
ames: more detailed "future msg" trace
...
This additional information may come in handy when debugging this crash
we shouldn't be seeing.
2022-08-03 20:46:54 +02:00
fang
e16c499724
ames: make recork timer faster
...
This will aid visibility and debugging. We'll probably want to make this
slower again later.
2022-08-03 20:33:04 +02:00
fang
7c01ee9b35
ames: account for corked flows during %stir
...
Also patches the behn scry in %stir handling to be correct for latest behn.
2022-08-03 19:37:53 +02:00
fang
b860bef066
ames: correctly avoid setting pump timer
...
The condition got butchered during refactor: instead of avoiding the creation
of pump timers during recork wake, it was setting them _exclusively_ during
recork wake.
2022-08-02 23:23:48 +02:00
Ted Blackman
e5e4c8602d
ames: properly set recork flag in +on-wake
2022-08-02 23:15:43 +03:00
Ted Blackman
67efb2ad7c
ames: don't start pump timers on recork
2022-08-02 23:03:34 +03:00
yosoyubik
a5aedaccc5
ames: delete naxplanation flows if bone is corked
2022-07-18 10:27:36 +02:00
yosoyubik
35c6ac416e
ames: fix predicate for detecting %cork acks
2022-07-16 07:40:05 +02:00
yosoyubik
c599e60c5b
ames: add recork set
2022-07-15 16:45:32 +02:00
yosoyubik
a3518ef4d6
ames: add trace for %cork plea
2022-07-13 09:17:42 +02:00
yosoyubik
34a7496720
ames: clean up nack flows after a cork
2022-07-08 07:29:04 +02:00
yosoyubik
ca1b586a46
ames: init cork timer on larval +take
2022-07-07 11:35:28 +02:00
yosoyubik
d4d443db13
ames: remove cork wire from state
2022-07-07 09:20:57 +02:00
yosoyubik
e366e55dbc
ames: add global daily timer for pending corks
...
WIP compiles and upgrades, and the timer gets properly reset
but untested for actual re-subscriptions with old publishers
2022-07-06 17:00:35 +02:00
yosoyubik
76919ce07c
ames: resend %cork after a naxplanation in one day
2022-07-06 10:52:19 +02:00
yosoyubik
ea4a31558f
ames: ack last fragment from closing bone
2022-07-01 09:59:24 +02:00
yosoyubik
a967b42c30
ames: print bone if "hear last in-progress"
2022-06-30 17:12:07 +02:00
yosoyubik
b0766c2f53
ames: remove misleading trace printf
2022-06-30 10:46:02 +02:00
yosoyubik
27ac0b601e
ames: use %$ as recipient vane for %corks
2022-06-30 10:18:37 +02:00
yosoyubik
17d52b8535
ames: handle bounce %cork from old publisher
...
Because the publisher will send the cork plea back to the subscriber on
the next bone, we are not able to know the bone for the original cork.
To handle it, we add the cork bone to the plea path
still wip: it keeps resending the cork plea faster than its ~h1 timer
2022-06-28 12:48:07 +02:00
yosoyubik
7de9c72f36
ames: add rift when inserting new peer state
2022-06-27 13:55:07 +02:00
yosoyubik
2e92331a0b
Merge branch 'next/arvo' into yu/gall-rq-wire-ames-flow-kill
2022-06-16 08:54:24 +02:00
Ted Blackman
0190a477f5
ames: adjust comments
2022-06-15 15:35:41 -06:00
Ted Blackman
1380e83b55
ames: clog on 5 messages and 100kB
2022-06-15 15:28:34 -06:00
Ted Blackman
d2ea33ddfe
ames: condense message size calculation
...
Co-authored-by: fang <git@fang.io>
2022-06-05 08:49:47 -05:00
Ted Blackman
953273004a
ames: clog at 100kB instead of 5 messages
2022-06-04 17:18:21 -05:00
fang
43dff9c279
ames: do not use old lanes during stateful forward
...
As 999f0e8
2022-06-01 21:47:15 +02:00
fang
999f0e84e9
ames: do not produce stale forward lanes
...
Stale lanes may cause forwarding loops. Imagine the following:
1) Planet A is live. Galaxy B, its indirect sponsor, learns of its route.
2) A goes offline. Another ship, C, is started in its place, at the same route.
3) B receives a packet for A, forwards it to the known route.
4) C received the packet, forwards it to B.
5) Repeat from 3.
Here, we update the forward lane(s) scry used by the runtime to not produce a
peer's lane if they haven't communicated with us in the last hour. Everyone's
supposed to ping their sponsorship chain every 30 seconds. If those aren't
going through, you shouldn't expect to be reachable anyway.
We may or may not want to update +send-blob to match.
2022-06-01 17:17:37 +02:00
Philip Monk
5b773723b3
Merge remote-tracking branch 'origin/yu/ames-fixes' into next/arvo
2022-05-20 22:52:00 -07:00
Philip Monk
ac64f44538
clay: update tests
2022-05-20 20:03:34 -07:00
Philip Monk
f16abbd992
ames: fix comet communication with breached ship
2022-05-19 20:20:00 -07:00
yosoyubik
16920a894e
Merge branch 'yu/ames-fixes' into yu/gall-rq-wire-ames-flow-kill
2022-05-18 17:37:24 +02:00
yosoyubik
c2576cdc46
ames: handle +call moves if re-evolving from larva
2022-05-18 17:36:40 +02:00
yosoyubik
fe412b1ca7
gall: remove $watches from egg/yoke
...
Also renames $beat to $boar so it signifies more its closeness to $boat
2022-05-18 15:27:17 +02:00
Philip Monk
7e4ad62a51
Merge remote-tracking branch 'origin/yu/ames-fixes' into philip/tomb
2022-05-17 14:07:05 -07:00
yosoyubik
4dae3ae3e2
ames: indent
2022-05-17 07:49:15 +02:00
yosoyubik
f81061fe40
Merge branch 'yu/ames-fixes' into yu/gall-cork-wip
2022-05-17 07:47:58 +02:00
yosoyubik
2553573be0
ames: start drainage timer if regressed from adult
...
This accounts for a possible race condition where ames expects a
response, but regresses into the larval state. Upon receiving the
$sign on +take, we would remain stuck as a larva. Now we check
that we have enough information to re-evolve and then start a
/larval timer to begin draining the queue.
2022-05-17 07:11:09 +02:00
Philip Monk
d25194c7a3
arvo: reduce priority of lifecycle printfs
2022-05-14 23:59:21 -07:00
Philip Monk
8a2eac7c70
Merge remote-tracking branch 'origin/ted/ames-prod-2' into philip/tomb
2022-05-14 21:17:10 -07:00
Philip Monk
2e31f87068
Merge branch 'pr' into philip/tomb
2022-05-14 20:59:01 -07:00
yosoyubik
350636fe2a
ames: don't relay acks for naxplanations of %corks
...
This avoids closing a subscription prematurely, before the
server ames has had time to process the innitial %cork.
2022-05-14 10:52:45 +02:00
yosoyubik
76ddc359f5
ames: fixt 6-to-7 state update
2022-05-09 14:17:27 +02:00
yosoyubik
a2b771f704
Merge branch 'yu/ames-fixes' into yu/gall-cork-wip
2022-05-09 14:16:07 +02:00
yosoyubik
5d6b2d8997
ames: fix 4-to-5 state migration
2022-05-09 14:13:51 +02:00
yosoyubik
a278b63c66
ames: add tracing printf
2022-05-06 09:47:19 +02:00
yosoyubik
6318aa283b
ames: retry sending a %cork if publisher is behind
2022-05-05 14:04:18 +02:00
yosoyubik
06883b617e
ames: add new state version in +stay
2022-05-05 13:00:23 +02:00
yosoyubik
53e1c86833
Merge remote-tracking branch 'origin/yu/ames-fixes' into yu/gall-rq-wire-ames-flow-kill
2022-05-03 13:57:39 +02:00
yosoyubik
ec60399cdf
ames: restore previous +decode-shut-packet
2022-05-03 12:49:10 +02:00
yosoyubik
3f7f4dbd5a
gall: handle %kicks for nonexistent subscriptions
2022-05-03 12:49:10 +02:00
yosoyubik
a288d455c3
ames: remove %cork from |ack-meat
2022-05-03 12:49:10 +02:00
yosoyubik
8be9257164
ames: style indentation
2022-05-03 12:49:09 +02:00
yosoyubik
52ef6ba08e
ames: handles decoding old-format shut-packets
2022-05-03 12:49:09 +02:00
h5gq3
c95cce4404
style fixes
2022-05-03 12:49:09 +02:00
h5gq3
5583fc1320
ames-flow-kill
2022-05-03 12:49:09 +02:00
yosoyubik
02806f63a1
ames: hardcode rift for comets in state upgrade
2022-04-27 06:24:55 +02:00
yosoyubik
9cb425e0ef
ames: remove unused ames-to-ames %plea logic
...
This seems to have been used in the past for pinging the ship's
sponsor, but was later removed when the :ping app was introduced
2022-04-22 15:49:15 +02:00
yosoyubik
087f24e931
ames: use +trace to print logs
2022-04-22 12:29:16 +02:00
yosoyubik
e2f8e21529
ames: remove sigwut
2022-04-22 12:26:53 +02:00
yosoyubik
471b9be22d
ames: fix emit of parsing old wire log
2022-04-22 12:26:24 +02:00
yosoyubik
91435a92ca
ames: don't drop events with old wires
...
Previously we were dropping events that used old
wires that lacked a rift in them. This seems a
bad behavior because we don't want to destroy a
flow that has not been processed by both ends.
Note: pending a fix to test-old-ames-wire
2022-04-22 12:26:14 +02:00
yosoyubik
6cdebea1af
ames: address review
2022-04-22 12:26:08 +02:00
yosoyubik
60cc44e961
ames: fix cached state in larval upgrade
2022-04-22 12:26:02 +02:00
yosoyubik
1b01455fe5
ames: fix dangling-bone error
2022-04-22 12:25:55 +02:00
lukechampine
8399b7b96c
ames: add resend timer for attestation requests
2022-04-04 00:19:55 -04:00
lukechampine
53a0c87cdb
ames: add sendkeys-packet
2022-03-02 00:56:57 -05:00
lukechampine
df6318acd0
ames: fix comet self-attestation handling
2022-03-02 00:52:48 -05:00
lukechampine
1044d42a15
ames: don't enqueue blobs sent to unknown comets
2022-03-02 00:46:18 -05:00
yosoyubik
3d943ec201
ames: ignore missing peer-state on-publ-sponsor
2022-01-18 20:34:32 +01:00
Ted Blackman
bbc80f9d45
zuse,ames: ^^sein:title
2021-09-25 16:52:53 -04:00
Ted Blackman
bce8dfd477
ames: pop the next packet, not the first packet again
2021-08-28 23:01:15 +03:00
