* swarm/api: fix file descriptor leak in NewTestSwarmServer
Swarm storage (localstore) was not closed. That resulted a
"too many open files" error if `TestClientUploadDownloadRawEncrypted`
was run with `-count 1000`.
* cmd/swarm: speed up StartNewNodes() by parallelization
Reduce cluster startup time from 13s to 7s.
* swarm/api: disable flaky TestClientUploadDownloadRawEncrypted with -race
* swarm/storage: disable flaky TestLDBStoreCollectGarbage (-race)
With race detection turned on the disabled cases often fail with:
"ldbstore_test.go:535: expected surplus chunk 150 to be missing, but got no error"
* cmd/swarm: fix process leak in TestACT and TestSwarmUp
Each test run we start 3 nodes, but we did not terminate them. So
those 3 nodes continued eating up 1.2GB (3.4GB with -race) after test
completion.
6b6c4d1c27 changed how we start clusters
to speed up tests. The changeset merged together test cases
and introduced a global cluster. But "forgot" about termination.
Let's get rid of "global cluster" so we have a clear owner of
termination (some time sacrifice), while leaving subtests to use the
same cluster.
* travis: remove verbose from Swarm race tests
By removing -v our output will be cleaner, but the Travis job still
won't be terminated - due to 'no output for 10 minutes' - as keepalive
.sh produces a log line every 5 minutes.
* travis: extend Swarm race detection to p2p subpackages
As p2p/protocols, p2p/simulations and p2p/testing packages mostly
belong to the Swarm team.
dummyHook's fields were concurrently written by nodes and read by
the test. The simplest solution is to protect all fields with a mutex.
Enable: TestMultiplePeersDropSelf, TestMultiplePeersDropOther as they
seemingly accidentally stayed disabled during a refactor/rewrite
since 1836366ac1.
resolvesethersphere/go-ethereum#1286
These tests never run as the build tag excluded them from the CI
execution. As a results the (dead) code got out of sync with other
parts of Swarm and now they would not even compile. => Removed.
resolvesethersphere/go-ethereum#1238
* swarm/storage: increase mget timeout in common_test.go
TestDbStoreCorrect_1k sometimes timed out with -race on Travis.
--- FAIL: TestDbStoreCorrect_1k (24.63s)
common_test.go:194: testStore failed: timed out after 10s
* swarm: remove unused vars from TestSnapshotSyncWithServer
nodeCount and chunkCount is returned from setupSim and those values
we use.
* swarm: move race/norace helpers from stream to testutil
As we will need to use the flag in other packages, too.
* swarm: refactor TestSwarmNetwork case
Extract long running test cases for better visibility.
* swarm/network: skip TestSyncingViaGlobalSync with -race
As panics on Travis.
panic: runtime error: invalid memory address or nil pointer dereference
[signal SIGSEGV: segmentation violation code=0x1 addr=0x0 pc=0x7e351b]
* swarm: run TestSwarmNetwork with fewer nodes with -race
As otherwise we always get test failure with `network_test.go:374:
context deadline exceeded` even with raised `Timeout`.
* swarm/network: run TestDeliveryFromNodes with fewer nodes with -race
Test on Travis times out with 8 or more nodes if -race flag is present.
* swarm/network: smaller node count for discovery tests with -race
TestDiscoveryPersistenceSimulationSimAdapters failed on Travis with
`-race` flag present. The failure was due to extensive memory usage,
coming from the CGO runtime. Using a smaller node count resolves the
issue.
=== RUN TestDiscoveryPersistenceSimulationSimAdapter
==7227==ERROR: ThreadSanitizer failed to allocate 0x80000 (524288) bytes of clock allocator (error code: 12)
FATAL: ThreadSanitizer CHECK failed: ./gotsan.cc:6976 "((0 && "unable to mmap")) != (0)" (0x0, 0x0)
FAIL github.com/ethereum/go-ethereum/swarm/network/simulations/discovery 804.826s
* swarm/network: run TestFileRetrieval with fewer nodes with -race
Otherwise we get a failure due to extensive memory usage, as the CGO
runtime cannot allocate more bytes.
=== RUN TestFileRetrieval
==7366==ERROR: ThreadSanitizer failed to allocate 0x80000 (524288) bytes of clock allocator (error code: 12)
FATAL: ThreadSanitizer CHECK failed: ./gotsan.cc:6976 "((0 && "unable to mmap")) != (0)" (0x0, 0x0)
FAIL github.com/ethereum/go-ethereum/swarm/network/stream 155.165s
* swarm/network: run TestRetrieval with fewer nodes with -race
Otherwise we get a failure due to extensive memory usage, as the CGO
runtime cannot allocate more bytes ("ThreadSanitizer failed to
allocate").
* swarm/network: skip flaky TestGetSubscriptionsRPC on Travis w/ -race
Test fails a lot with something like:
streamer_test.go:1332: Real subscriptions and expected amount don't match; real: 0, expected: 20
* swarm/storage: skip TestDB_SubscribePull* tests on Travis w/ -race
Travis just hangs...
ok github.com/ethereum/go-ethereum/swarm/storage/feed/lookup 1.307s
keepalive
keepalive
keepalive
or panics after a while.
Without these tests the race detector job is now stable. Let's
invetigate these tests in a separate issue:
https://github.com/ethersphere/go-ethereum/issues/1245
* swarm/network: DRY out repeated giga comment
I not necessarily agree with the way we wait for event propagation.
But I truly disagree with having duplicated giga comments.
* p2p/simulations: encapsulate Node.Up field so we avoid data races
The Node.Up field was accessed concurrently without "proper" locking.
There was a lock on Network and that was used sometimes to access
the field. Other times the locking was missed and we had
a data race.
For example: https://github.com/ethereum/go-ethereum/pull/18464
The case above was solved, but there were still intermittent/hard to
reproduce races. So let's solve the issue permanently.
resolves: ethersphere/go-ethereum#1146
* p2p/simulations: fix unmarshal of simulations.Node
Making Node.Up field private in 13292ee897e345045fbfab3bda23a77589a271c1
broke TestHTTPNetwork and TestHTTPSnapshot. Because the default
UnmarshalJSON does not handle unexported fields.
Important: The fix is partial and not proper to my taste. But I cut
scope as I think the fix may require a change to the current
serialization format. New ticket:
https://github.com/ethersphere/go-ethereum/issues/1177
* p2p/simulations: Add a sanity test case for Node.Config UnmarshalJSON
* p2p/simulations: revert back to defer Unlock() pattern for Network
It's a good patten to call `defer Unlock()` right after `Lock()` so
(new) error cases won't miss to unlock. Let's get back to that pattern.
The patten was abandoned in 85a79b3ad3,
while fixing a data race. That data race does not exist anymore,
since the Node.Up field got hidden behind its own lock.
* p2p/simulations: consistent naming for test providers Node.UnmarshalJSON
* p2p/simulations: remove JSON annotation from private fields of Node
As unexported fields are not serialized.
* p2p/simulations: fix deadlock in Network.GetRandomDownNode()
Problem: GetRandomDownNode() locks -> getDownNodeIDs() ->
GetNodes() tries to lock -> deadlock
On Network type, unexported functions must assume that `net.lock`
is already acquired and should not call exported functions which
might try to lock again.
* p2p/simulations: ensure method conformity for Network
Connect* methods were moved to p2p/simulations.Network from
swarm/network/simulation. However these new methods did not follow
the pattern of Network methods, i.e., all exported method locks
the whole Network either for read or write.
* p2p/simulations: fix deadlock during network shutdown
`TestDiscoveryPersistenceSimulationSimAdapter` often got into deadlock.
The execution was stuck on two locks, i.e, `Kademlia.lock` and
`p2p/simulations.Network.lock`. Usually the test got stuck once in each
20 executions with high confidence.
`Kademlia` was stuck in `Kademlia.EachAddr()` and `Network` in
`Network.Stop()`.
Solution: in `Network.Stop()` `net.lock` must be released before
calling `node.Stop()` as stopping a node (somehow - I did not find
the exact code path) causes `Network.InitConn()` to be called from
`Kademlia.SuggestPeer()` and that blocks on `net.lock`.
Related ticket: https://github.com/ethersphere/go-ethereum/issues/1223
* swarm/state: simplify if statement in DBStore.Put()
* p2p/simulations: remove faulty godoc from private function
The comment started with the wrong method name.
The method is simple and self explanatory. Also, it's private.
=> Let's just remove the comment.
* swarm/network: fix data race in stream.(*Peer).handleOfferedHashesMsg()
handleOfferedHashesMsg() contained a data race:
- read => in a goroutine, call to c.batchDone()
- write => in the main thread, write to c.sessionAt
c.batchDone() contained a call to c.AddInterval(). Client was a value
receiver for AddInterval. So on c.AddInterval() call the whole client
struct got copied (read) while one of its field was modified in
handleOfferedHashesMsg() (write).
fixesethersphere/go-ethereum#1086
* swarm/network: simplify some trivial statements
- Have `${DataDir}/bzzd.ipc` as IPC path default.
- Respect the `--datadir` flag.
- Keep only the global `--ipcpath` flag and drop the local `--ipcpath` flag
as flags might overwrite each other. (Note: before global `--ipcpath`
was ignored even if it was set)
fixes ethersphere#795
On file access LDBStore's tryAccessIdx() function created a faulty
GC Index Data entry, because not indexing the ikey correctly.
That caused the chunk addresses/hashes to start with '00' and the last
two digits were dropped. => Incorrect chunk address.
Besides the fix, the commit also contains a schema change which will
run the CleanGCIndex() function to clean the GC index from erroneous
entries.
Note: CleanGCIndex() rebuilds the index from scratch which can take
a really-really long time with a huge DB (possibly an hour).
Access count was not incremented when chunk was retrieved
from cache. So the garbage collector might have deleted the most
frequently accessed chunk from disk.
Co-authored-by: Ferenc Szabo <ferenc.szabo@ethereum.org>
* swarm: clean up unused private types and functions
Those that were identified by code inspection tool.
* swarm/storage: move/add Proximity GoDoc from deleted private function
The mentioned proximity() private function was deleted in:
1ca8fc1e6fa0ab4ab1aaca06d6fb32e173cd5f2f
When CLI tests were spanning new nodes, the log level verbosity was
hard coded as 6. So the Swarm process was always polluting the test
output with TRACE level logs.
Now `go test -v ./cmd/swarm -loglevel 0` works as expected.
Ferenc Szabo
