fang
ab75e83398
gall: don't enqueue %cork as outstanding
...
Ames will no longer send acks. Consider it completed immediately, and
clean up the outstanding queue whenever it's empty.
2022-08-25 18:14:40 +02:00
fang
8108595b2b
gall: don't inject potentially-fake acks
...
We shouldn't be hitting this case anymore. Injecting events into agents
which they might not expect, and might not even be real, is bad.
2022-08-19 21:39:44 +02:00
fang
59f0556e3b
Merge pull request #5958 from urbit/ted/rake
...
gall: %rake task to clean up dead %leave's
2022-08-19 21:29:08 +02:00
fang
a9a8619daa
Merge pull request #5947 from urbit/ted/staying-alive-rebase
...
gall: revive %hood and %dojo on keypress
2022-08-19 21:22:28 +02:00
fang
cfa18b1f1f
gall: when kicked, register %cork as outstanding
...
When ames notifies us that our subscription has been kicked, we enqueue
a cork to clean up the flow. Unlike the %leave case, however, we were
not registering the cork in the queue of outstanding comms. We would
eventually get an ack, but not know what for, and erroneously inject
%poke-acks and %watch-acks.
Here we simply add a %cork entry to the queue before sending it.
2022-08-19 16:04:36 +02:00
fang
c8a163fedb
gall: clean up big wire construction
2022-08-18 22:46:58 +02:00
fang
a5335f5851
gall: improve match detection in +ap-rake
...
Making sure to stringify the nonce and including all the target details in the
wire, and excluding flows that are already closing.
2022-08-18 22:39:54 +02:00
Ted Blackman
d8df2787a0
gall: %rake task to clean up dead %leave's
2022-08-18 20:34:42 +03:00
fang
18019d6d83
gall: re-enable doff, only for old-style subs
...
This is sufficient to bring the normal (non-prerelease-bugged) cases
into the new world.
For the prerelease ships that ran a buggier version of the new gall
subscription logic, we note that the conditional may trigger for the
nonce=1 case where it had already triggered for their
(shouldn't-be-possible) nonce=0 case. This results in a %leave on a wire
that wasn't in use. This no-ops on the publisher side though, and the
flow gets corked right away, so this is considered harmless.
2022-08-16 23:02:59 +02:00
Ted Blackman
c972c0fbe2
gall: roundtrip through arvo to revive %hood and %dojo
2022-08-16 01:37:35 +03:00
Ted Blackman
95580f7e83
gall: revive %hood and %dojo on poke
2022-08-16 00:40:38 +03:00
Ted Blackman
a70c2ffe1a
gall: now restarts %hood, but not %dojo yet
2022-08-16 00:40:38 +03:00
Ted Blackman
ac072729c2
gall: force load %hood and %dojo on poke
2022-08-16 00:40:38 +03:00
fang
81f18d13ad
gall: remove outdated fixme comment
...
This is part of the %doff work, which is getting integrated elsewhere.
2022-08-15 12:24:18 +02:00
Ted Blackman
1898483a45
gall: disable %doff for now
2022-08-12 00:55:19 +03:00
Ted Blackman
b1bab79f58
gall: add %f scry for sub-nonce's
2022-08-11 22:33:57 +03:00
fang
350f33d6ba
Merge branch 'next/arvo' into ted/doff-6
2022-08-11 21:06:30 +02:00
fang
5b4a964cdf
gall: fix the %doff conditional
...
No comment.
2022-08-11 19:48:27 +02:00
fang
06b7f52755
gall: print less during %doff
2022-08-11 17:04:44 +02:00
fang
cd0ea06ad4
gall: use a fully-formed wire when sending %cork
...
In response to clog notification from remote ames, we were sending a
%cork to clean up the flow. However, the wire we were using had the /sys
prefix already stripped off. Here, we put it back in.
2022-08-11 16:53:45 +02:00
fang
39cd895a55
gall: in +ap-doff, process oldest subs first
...
Start by killing subscription nonce 0, then work our way up instead of
down. We enhance the printf with a "total nonces" indicator so we can
still easily see the progress being made.
2022-08-10 22:51:38 +02:00
fang
2fe27bd8d7
gall: simplify +ap-doff filter conditional
2022-08-10 22:25:01 +02:00
fang
04b7b86cae
gall: support %doff-ing specific apps and/or ships
...
This gives us more control over exactly which subscriptions get culled,
which is useful in debugging and targetted recovery scenarios.
2022-08-10 20:33:07 +02:00
Ted Blackman
9e47490c95
gall: fix %doff handling
...
Previous +ap-doff kicked the agent repeatedly. We needed to kick
it only once. Now publisher agents clear their incoming subscription
state without the subscriber making lots of new subscriptions because
of repeated kicking.
2022-08-10 19:48:39 +03:00
Ted Blackman
ef09595788
gall: update comment
2022-08-10 19:13:32 +03:00
Ted Blackman
543efac183
gall: kill old subscriptions on %doff (still has bug)
...
Squashed commit of the following:
commit 1e31ab96c14e7506d5cb746d90153cb5658df6e0
Author: Ted Blackman <ted@tlon.io>
Date: Wed Aug 10 19:08:39 2022 +0300
ames: revert prints on bad memos
commit c47f9c6053c05da0177f0c64210a88161e8adc98
Author: Ted Blackman <ted@tlon.io>
Date: Wed Aug 10 18:44:56 2022 +0300
Merge fix to +ap-kill-down and use that pattern in +mo-doff.
This doesn't seem to work properly, but I'm not sure why so far.
commit e696816b8d3ab81534de6c2910d0fd4bebcfb68a
Merge: 4edb247cd7 a5f08643ff8eb8a1da29aad5fa6fae731e27d5a1
2022-08-10 19:09:47 +03:00
fang
682dde904c
Merge pull request #5931 from urbit/m/gall-kill-down-accurately
...
gall: handle nonces in wires correctly
2022-08-10 00:04:10 +02:00
fang
db1e8d1b92
gall: explain reason for strange conditional
2022-08-09 21:36:08 +02:00
fang
52d8ef6a99
gall: initialize sub-nonce as 1 during upgrade
...
This matches the bunt of 1 for the new ship case. A nonce of 0 is
semantically mapped to old-style subscriptions.
2022-08-09 21:22:40 +02:00
fang
c2d77a5d47
gall: in/exclude the nonce from the wire as needed
...
+ap-nuke was not including the nonce, but should.
+ap-handle-peers was potentially including a zero nonce.
(The latter shouldn't have been possible, but there's a bug in +load
where sub-nonce.yoke gets initialized as 0 instead of 1.)
2022-08-09 21:21:10 +02:00
fang
0a3f542fcc
gall: refactor sub-wire-with-nonce construction
...
Into +ap-nonce-wire.
2022-08-09 21:20:24 +02:00
fang
1f30dccddf
gall: use correct wires in +ap-kill-down
...
For potentially-outgoing moves, we need to make sure the nonce is in the
wire so that it gets associated with the right instance of the
subscription.
2022-08-09 20:53:43 +02:00
fang
e396ddcdaa
gall: cork a subscription's ames flow on-kick
...
Gall tells ames to %cork flows for subscriptions it has closed.
Receiving a kick also closes a subscription, but gall wasn't issuing a
%cork in that case. We correct that here.
Inlines +mo-handle-ames-response's logic at its only callsite.
2022-08-08 22:21:34 +02:00
yosoyubik
c69af12a6d
gall: make spore-tag a @ud
...
TODO: still getting kiln: %base not installed
2022-07-18 13:49:16 +02:00
yosoyubik
8712804d7f
gall: comment subscription kill
2022-07-14 07:23:39 +02:00
yosoyubik
d17fd7521a
gall: use spore tag for +molt subscription killing
...
WIP: kiln crashes after upgrade with a ! kiln: %base not installed
2022-07-13 15:52:15 +02:00
Ted Blackman
1cb2aaabbc
gall: fix +molt initialization
2022-07-13 15:52:07 +02:00
Ted Blackman
5f8378e1b7
gall: kill subscriptions after agents reload
2022-07-13 15:52:02 +02:00
yosoyubik
d9b7643b79
gall: fix condition to run +ap-kill-down
2022-07-13 15:51:46 +02:00
yosoyubik
4c8a4ecb0e
gall: run +ap-kill-down only once
2022-07-13 15:51:38 +02:00
yosoyubik
354a914c82
gall: run +ap-kill-down on +molt
2022-07-13 15:51:28 +02:00
yosoyubik
608aba2a87
gall: handle subscriber-side %cork
...
this also enables increasing sub-nonces for every new subscription
TODO: create one global cork timer for all peers
2022-07-06 14:44:20 +02:00
yosoyubik
f3a60aef14
gall: don't send %cork when leaving a subscription
...
This is a temporary fix, and first part of the gall-request-queue-fix
release in two stages. This gives a publisher ship the ability to
understand a %cork and handle it properly, but no subscriber will
be sending %corks at this stage when leaving a subscription.
We still add a nonce to all subscription wires but it doesn't
increment it when resubscribing, allowing flows to be reused.
Tested locally with toy pub/sub agents and Group join/leaving
2022-07-06 10:59:08 +02:00
yosoyubik
850de0bfd3
Revert "gall: run +ap-kill-down on +molt"
...
This reverts commit 2107240962
2022-07-06 09:12:30 +02:00
yosoyubik
675cf45171
Revert "gall: run +ap-kill-down only once"
...
This reverts commit cb5893ac05
2022-07-06 09:12:10 +02:00
yosoyubik
3560f6b7e3
Revert "gall: fix condition to run +ap-kill-down"
...
This reverts commit e5c86886e4
2022-07-06 09:11:22 +02:00
yosoyubik
e5c86886e4
gall: fix condition to run +ap-kill-down
2022-06-30 17:44:11 +02:00
yosoyubik
cb5893ac05
gall: run +ap-kill-down only once
2022-06-30 17:06:22 +02:00
yosoyubik
2107240962
gall: run +ap-kill-down on +molt
2022-06-30 15:57:02 +02:00
yosoyubik
2e92331a0b
Merge branch 'next/arvo' into yu/gall-rq-wire-ames-flow-kill
2022-06-16 08:54:24 +02:00
