Updates the package and fixes a test that depended on the old behavior
of Write() being equivalent to Reset()+Write() which is no longer the
case. The scanner already did resets after each block write, so this is
fine.
Given that we've taken on the resposibility of maintaining this forked
package I've added it to the Syncthing organization. We still vendor it
like an external package, because it's convenient to keep it as a fork
of upstream to easier merge and file pull requests towards them.
This is a new revision of the discovery server. Relevant changes and
non-changes:
- Protocol towards clients is unchanged.
- Recommended large scale design is still to be deployed nehind nginx (I
tested, and it's still a lot faster at terminating TLS).
- Database backend is leveldb again, only. It scales enough, is easy to
setup, and we don't need any backend to take care of.
- Server supports replication. This is a simple TCP channel - protect it
with a firewall when deploying over the internet. (We deploy this within
the same datacenter, and with firewall.) Any incoming client announces
are sent over the replication channel(s) to other peer discosrvs.
Incoming replication changes are applied to the database as if they came
from clients, but without the TLS/certificate overhead.
- Metrics are exposed using the prometheus library, when enabled.
- The database values and replication protocol is protobuf, because JSON
was quite CPU intensive when I tried that and benchmarked it.
- The "Retry-After" value for failed lookups gets slowly increased from
a default of 120 seconds, by 5 seconds for each failed lookup,
independently by each discosrv. This lowers the query load over time for
clients that are never seen. The Retry-After maxes out at 3600 after a
couple of weeks of this increase. The number of failed lookups is
stored in the database, now and then (avoiding making each lookup a
database put).
All in all this means clients can be pointed towards a cluster using
just multiple A / AAAA records to gain both load sharing and redundancy
(if one is down, clients will talk to the remaining ones).
GitHub-Pull-Request: https://github.com/syncthing/syncthing/pull/4648
These functions were very naive and slow. We haven't done much about
them because they pretty much don't matter at all for Syncthing
performance. They are however called very often in the discovery server
and these optimizations have a huge effect on the CPU load on the
public discovery servers.
The code isn't exactly obvious, but we have good test coverage on all
these functions.
benchmark old ns/op new ns/op delta
BenchmarkLuhnify-8 12458 1045 -91.61%
BenchmarkUnluhnify-8 12598 1074 -91.47%
BenchmarkChunkify-8 10792 104 -99.04%
benchmark old allocs new allocs delta
BenchmarkLuhnify-8 18 1 -94.44%
BenchmarkUnluhnify-8 18 1 -94.44%
BenchmarkChunkify-8 44 2 -95.45%
benchmark old bytes new bytes delta
BenchmarkLuhnify-8 1278 64 -94.99%
BenchmarkUnluhnify-8 1278 64 -94.99%
BenchmarkChunkify-8 42552 128 -99.70%
GitHub-Pull-Request: https://github.com/syncthing/syncthing/pull/4346
This updates kcp and uses our own fork which:
1. Keys sessions not just by remote address, but by remote address +
conversation id 2. Allows not to close connections that were passed directly
to the library. 3. Resets cache key if the session gets terminated.
GitHub-Pull-Request: https://github.com/syncthing/syncthing/pull/4339
LGTM: calmh
Can't do what I did, as the rolling function is not the same as the
non-rolling one. Instead this uses an improved version of the rolling
adler32 to accomplish the same thing. (PR filed on upstream, so should
be able to use that directly in the future.)
