8cc3e4d3c6
Merge release-6.3 into master
2020-10-19 22:51:56 -07:00
a9607bdcec
Explicitly seal classes that inherit but aren't inherited from
2020-10-07 21:58:24 -07:00
8571dcfe28
Use override where applicable in fdbserver
2020-10-07 18:41:19 -07:00
2c5a2909cd
fixed compiler error
2020-10-06 09:44:16 -07:00
59d4e8b629
added a timeout to TLS handshakes
2020-10-05 15:24:03 -07:00
0ac08f6a9b
Replace NULL with nullptr in flow
2020-09-20 11:31:49 -07:00
a49cb41de7
Merge branch 'release-6.3'
...
# Conflicts:
# CMakeLists.txt
# cmake/ConfigureCompiler.cmake
# fdbserver/Knobs.cpp
# fdbserver/StorageCache.actor.cpp
# fdbserver/storageserver.actor.cpp
# flow/ThreadHelper.actor.h
# flow/serialize.h
# tests/CMakeLists.txt
2020-07-29 00:31:55 -07:00
c4f3553826
TLS handshake limit changed to regular flow lock, added new knob for simultaneous handshakes limit and removed old knobs. Added knob to control how much data the FlowTransport connectionWriter() can write using multiple write() calls before yielding to other tasks, and updated p2p network test to use the same logic. Changed knob defaults for send limit on single writes and TLS handshake thread count.
2020-07-16 02:31:17 -07:00
5151f1167c
Added TLS option to limit the number of glibc malloc arenas if TLS background threads are created. TLS initialization now has three phases in order to avoid creating TLS threads or limit the malloc arena count unnecessarily.
2020-06-28 17:29:13 -07:00
813dbfb297
TLS handshake threads are now created with TLS initialization instead of when the network is created since they may not be needed and it avoids calling startThread() with a custom stack size too early in process execution. StartThread() now logs a warning if the given stack size can't be used. Reorganized TLS handshake thread state near the other TLS related stuff.
2020-06-23 03:09:56 -07:00
c3ce0034bf
Background threads for TLS handshakes use a stack size based on a knob. Platform::startThread() now accepts a stack size. The generic thread pool implementation takes an optional stack size override which is used for each added thread.
2020-06-23 02:08:01 -07:00
5b19233989
Implemented count of used background TLS handshake threads differently using a generic utility class that updates the count on destruction.
2020-06-23 00:45:07 -07:00
a48d005f61
Bug fixes, stats improvements.
2020-06-22 13:37:13 -07:00
d75b4c9a69
Bug fix, accidental local variable redeclaration.
2020-06-22 05:53:37 -07:00
d681e9d350
Follow TLS_DISABLED macro setting.
2020-06-21 11:21:24 -07:00
6330f8b458
TLS handshakes can now be done using up to a configured number of background threads if any of them are not busy.
2020-06-21 06:23:32 -07:00
d7cb54d2b2
Fixed Windows CMake build
2020-06-20 23:03:18 -07:00
a59925dd73
Added more const-correctness improvements
2020-06-20 22:15:19 -07:00
99bf993815
Replace BOOST_NOEXCEPT with noexcept
2020-06-09 22:39:19 -07:00
30bfd606c0
Merge branch 'release-6.2' into release-6.3
...
# Conflicts:
# CMakeLists.txt
# documentation/sphinx/source/downloads.rst
# documentation/sphinx/source/release-notes.rst
# fdbserver/worker.actor.cpp
# packaging/msi/FDBInstaller.wxs
# versions.target
2020-06-04 19:21:32 -07:00
bf072d68ec
Workarounds for strange behaviors in Boost ssl sockets on MacOS and Linux. When writing to the ssl socket, write_some() would sometimes return BrokenPipe instead of WouldBlock unless onWriteable on the raw socket was checked first. On MacOS, even with the onWriteable check using a send size greater than 2016 (determined experimentally) would still result in the error. Also consolidated two identical copies of SendBufferIterator.
2020-06-04 01:44:05 -07:00
86f712657e
Eliminate the undefined behavior of calling run_network twice, instead returning an error.
2020-05-22 13:31:06 -07:00
c2e80fe47b
Linux aarch64 port
2020-05-09 22:14:03 +00:00
95bc24de11
flow: update headers and includes
2020-04-30 18:11:23 +00:00
c87aa33941
Merge branch 'release-6.2'
...
# Conflicts:
# CMakeLists.txt
# bindings/go/src/fdb/generated.go
# documentation/sphinx/source/api-common.rst.inc
# documentation/sphinx/source/api-ruby.rst
# documentation/sphinx/source/release-notes.rst
# fdbclient/FailureMonitorClient.actor.cpp
# fdbclient/NativeAPI.actor.cpp
# fdbclient/vexillographer/fdb.options
# fdbrpc/FlowTransport.actor.cpp
# fdbserver/OldTLogServer_6_0.actor.cpp
# fdbserver/TLogServer.actor.cpp
# fdbserver/fdbserver.actor.cpp
# versions.target
2020-04-23 13:47:53 -07:00
c6df20a179
Use nullptr instead of NULL
2020-04-21 20:39:45 -07:00
a51746b307
Match 6.2.15's behavior in how invalid/unreadable/non-existent certs are handled.
...
Which is to proceed past Net2 creation, and allow certificate refresh to
try and eventually load valid certs. Additionally, fix certificate
refeshing dieing if the certificate is not readable when first called.
In testing, I also found and fixed an issue where if a cert went from
unreadable to readable, we wouldn't reload the TLS context, due to not
considering it as a file change.
2020-04-20 21:38:04 -07:00
20fe068863
Merge branch 'tls-background-eio-thread' into tls-permission-errors
2020-04-20 20:51:05 -07:00
49c6bb90ef
Merge pull request #2982 from alexmiller-apple/tls-log-settings
...
Log Net2TLSConfig with paths and settings when using TLS.
2020-04-20 15:46:26 -07:00
75a4f3b7c9
Remove comment about ignoring runOnMainThread errors.
...
If we got an exception, it wouldn't be of type `Error` anyway, so
it seems like things would crash regardless.
2020-04-20 13:19:42 -07:00
da8e47ea25
Merge remote-tracking branch 'upstream/release-6.2' into tls-background-eio-thread
2020-04-20 13:15:05 -07:00
5c399bf725
Move the callbacks into ::run() right before it exits.
...
stopped=true doesn't cause the run loop to immediately exit.
2020-04-20 13:14:19 -07:00
2ce539ef6d
Respect flow<->fdbrpc module boundaries.
...
Which fixes a compilation error due to a circular dependency between
flow.a and fdbrpc.a. However, this is now done at the cost of newNet2
users have to remember to add Net2FileSystem::stop() as a callback.
2020-04-20 02:53:07 -07:00
cbb6ffb431
Only log OpenSSL error strings for OpenSSL errors.
...
Normal "connection refused" messages would show up with a long verbose
string that doesn't really provide any useful information otherwise.
2020-04-18 20:39:02 -07:00
11eebc4a48
Log Net2TLSConfig with paths and settings when using TLS.
...
There were similar TraceEvents in the FDBLibTLS/LibreSSL TLS
implementaiton that were accidentally dropped in the TLS rewrite.
This makes it so that one does not have to use magic to figure out if a
process was configued with TLS correctly when some of the settings come
from environment variables.
2020-04-18 20:21:10 -07:00
1398e9a82e
Stop background eio threads on Net2::stop().
...
This will stop eio threads for both the client (`fdb_stop_network()`)
and the server. This change is being done more for the former, but I
don't see any harm in doing the latter as well.
2020-04-18 19:40:55 -07:00
d8690d31cd
Merge branch 'master' into per-priority-busy-logging
...
# Conflicts:
# flow/Net2.actor.cpp
2020-04-15 08:31:30 -07:00
b1172417f5
Merge branch 'master' into per-priority-busy-logging
...
# Conflicts:
# flow/Knobs.cpp
# flow/Knobs.h
# flow/Net2.actor.cpp
2020-04-14 14:22:12 -07:00
e104a2e3a6
Merge commit 'cf01233f28a2c42908656a39f458a4475c1d44a3' into run-loop-busy-profiler
...
# Conflicts:
# documentation/sphinx/source/release-notes.rst
# fdbclient/NativeAPI.actor.h
# fdbserver/fdbserver.actor.cpp
# flow/Net2.actor.cpp
2020-04-14 14:02:24 -07:00
cf01233f28
Merge pull request #2549 from ajbeamon/slow-task-and-priority-tracking-improvements
...
Slow task and priority tracker improvements
2020-04-14 13:26:32 -07:00
903128a36f
Fix boost error trace event fields to match other trace events by the same name.
2020-04-01 08:18:50 -07:00
e08f0201f1
merge release 6.2 into master
2020-03-17 12:51:47 -07:00
db5863145a
Merge remote-tracking branch 'upstream/release-6.2' into fdbcli-tlsinfo
2020-03-16 12:33:50 -07:00
75e2fffe5a
Add a ProcessMetrics.TLSPolicyFailures metric
...
This reports the number of policy failures over the past 5s interval.
It also is step 1 towards getting this information into status json.
2020-03-13 02:24:37 -07:00
0c558efcfe
Add a `tlsinfo` command to fdbcli that prints the certificate chain.
...
This requires the certificate chain to load successfully, otherwise
fdbcli will error out at an earlier point due to Net2 not being able to
configure TLS.
2020-03-13 00:11:53 -07:00
303df197cf
Merge branch 'release-6.2'
...
# Conflicts:
# CMakeLists.txt
# bindings/c/test/mako/mako.c
# documentation/sphinx/source/release-notes.rst
# fdbbackup/backup.actor.cpp
# fdbclient/NativeAPI.actor.cpp
# fdbclient/NativeAPI.actor.h
# fdbserver/DataDistributionQueue.actor.cpp
# fdbserver/Knobs.cpp
# fdbserver/Knobs.h
# fdbserver/LogRouter.actor.cpp
# fdbserver/SkipList.cpp
# fdbserver/fdbserver.actor.cpp
# flow/CMakeLists.txt
# flow/Knobs.cpp
# flow/Knobs.h
# flow/flow.vcxproj
# flow/flow.vcxproj.filters
# versions.target
2020-03-06 18:22:46 -08:00
188d9b8239
Don't swallow actor cancellation in certificate refreshing.
2020-03-06 11:09:17 -08:00
9b760fae2d
Rewrite all Errors into tls_errors if they happen as part of initializing TLS.
2020-03-06 11:06:19 -08:00
ac52b6b474
Rework a bit of error and exception handling.
...
I went back and dug through all of the "what functions can throw what
types", and made sane decisions about them. boost errors are
aggressively translated into FDB ones, whcih might result in multiple
lines of logging about errors, but this is in infrequently run code, so
it should be fine.
2020-03-06 02:33:16 -08:00
ccef3f7d05
Attempt to fix TLS_DISABLED compiles.
2020-03-05 17:32:10 -08:00
Young Liu