GetRandomStripe function to randomly select a segment stripe to
audit was using `segment.EncryptedSize/segment.Redundancy.StripeSize()`.
Since integer divsion truncates it leads to skipping last stripe if
its size is less than stripe size. Use `Redundancy.StripeCount` to
get correct stripe count.
Change-Id: Ida09e035be30a21219ab3e1aedd66af8be707d1b
If we encounter an error during the infectious error correction, we just
add it to the errlist to be logged at the worker level.
We want to make sure we know about this if it happens. Give it its own
error log and increment a monkit metric.
Change-Id: Ie5946ae3cd97b766e3099af8ce160a686135ee27
When a node gets enough timeouts, it is supposed to be removed
from pending_audits and get an audit failure. We would give them
a failure, but we missed the removal. This change fixes it.
Change-Id: I2f7014e28d7d9b01a9d051f5bbb4f67c86c7b36b
"audit failed" is already used when a node fails an audit. That makes
searching for this higher level audit worker error more difficult.
Additionally, the presence of errors from the audit worker doesn't
necessarily mean the audit failed. Reword the error message to
"error(s) during audit"
Change-Id: I0aab12c73c18d4bd962c5d8ac8a17cabcec022e6
When we can't complete an audit or repair, we need more information about
what happened during each individual share/piece download.
In audit, add the number of offline, unknown, contained, failed nodes to
the error log. In repair, combine the errors from each download and add
them to the error log.
Change-Id: Ic5d2a0f3f291f26cb82662bfb37355dd2b5c89ba
package in audit
This PR implements reputation store and replace overlay in audit service
to use such store for storing node's audit stats.
In order to keep the changeset smaller, most of the changes in this PR is for copying audit logic in overlay to
reputation package. In a following PR, the duplicating code will be
removed from overlay.
Change-Id: I16c12494a0970f44c422b26cf603c1dc489e5bc1
In most situations where we would not get enough shares to complete
an audit, something has probably gone wrong like a forgotten delete,
and nodes should not be failed. We have an alert when this occurs.
Check the logs to see what happened. If we decide the nodes should
get audit failures, we can do it manually.
Change-Id: Ib6e408082048d31197c37ebfd7f9031135fc938f
Currently, pending audit is finding segment by segment location
(path) because we want to move audit to segmentloop and we will
have only StreamID and Position we need to add columns for those
fields. Altering existing table can cause issues while
migration and deployment. Cleaner choise is to make new table.
This change contains migration with new segment_pending_audit
table that will replace pending_audits table and adjustments
to use new table in the code.
Table pending_audits will be dropped with next release.
Change-Id: Id507e29c152da594bac1fd812c78d7ecf45ec51f
Satellites set their configuration values to default values using
cfgstruct, however, it turns out our tests don't test these values
at all! Instead, they have a completely separate definition system
that is easy to forget about.
As is to be expected, these values have drifted, and it appears
in a few cases test planet is testing unreasonable values that we
won't see in production, or perhaps worse, features enabled in
production were missed and weren't enabled in testplanet.
This change makes it so all values are configured the same,
systematic way, so it's easy to see when test values are different
than dev values or release values, and it's less hard to forget
to enable features in testplanet.
In terms of reviewing, this change should be actually fairly
easy to review, considering private/testplanet/satellite.go keeps
the current config system and the new one and confirms that they
result in identical configurations, so you can be certain that
nothing was missed and the config is all correct.
You can also check the config lock to see what actual config
values changed.
Change-Id: I6715d0794887f577e21742afcf56fd2b9d12170e
Until now, whenever audits were recorded we would try to delete
the node from containment just in case it exists. Since we now
want to treat segment repair downloads as audits, this would
erroneously remove nodes from containment, as repair does not go
through a Reverify step. With this changeset, (Batch)UpdateStats
will not remove nodes from containment. The Reverify method will
remove all necessary nodes from containment.
Change-Id: Iabc9496293076dccba32ddfa028e92580b26167f
Previously the object range was not used for calculating order limit.
This meant that even if you were downloading only a small range it would
account bandwidth based on the full segment.
This doesn't fully address the accounting since the lazy segment
downloads do not send their requested range nor requested limit.
Change-Id: Ic811e570c889be87bac4293547d6537a255078da
Currently the interface is not useful. When we need to vary the
implementation for testing purposes we can introduce a local interface
for the service/chore that needs it, rather than using the large api.
Unfortunately, this requires adding a cleanup callback for tests, there
might be a better solution to this problem.
Change-Id: I079fe4dbe297b0ae08c10081a1cea4dfbc277682
When audits are being recorded, we automatically add some SQL to remove
the node from the pending audits table in case it exists. They are
removed from pending audits even if the node was offline for the audit.
This is not the correct behavior.
Add statement to record audit results in reverify tests to ensure no
more false positives.
Change-Id: I186ae68bc5e7962ef6c5defbebc1d95e63596a17
errs.Class should not contain "error" in the name, since that causes a
lot of stutter in the error logs. As an example a log line could end up
looking like:
ERROR node stats service error: satellitedbs error: node stats database error: no rows
Whereas something like:
ERROR nodestats service: satellitedbs: nodestatsdb: no rows
Would contain all the necessary information without the stutter.
Change-Id: I7b7cb7e592ebab4bcfadc1eef11122584d2b20e0
Currently the loop handling is heavily related to the metabase rather
than metainfo.
metainfo over time has become related to the "public API" for accessing
the metabase data.
Currently updates monkit.lock, because monkit monitoring does not handle
ScopeNamed correctly. Needs a followup change to monitoring check.
Change-Id: Ie50519991d718dfb872ec9a0176a82e732c97584
metabase has become a central concept and it's more suitable for it to
be directly nested under satellite rather than being part of metainfo.
metainfo is going to be the "endpoint" logic for handling requests.
Change-Id: I53770d6761ac1e9a1283b5aa68f471b21e784198
Check that the bloom filter creation date is earlier than the
metainfo loop system time used for db scanning.
Change-Id: Ib0f47c124f5651deae0fd7e7996abcdcaac98fb4
At some point we might try to change original segment RS values and set Pieces according to the new values. This change adds add NewRedundancy parameter for UpdateSegmentPieces method to give ability to do that. As a part of change NewPieces are validated against NewRedundancy.
Change-Id: I8ea531c9060b5cd283d3bf4f6e4c320099dd5576
Among other conditions, nodes fail audits by returning incorrect
data and by reaching the max reverify count, but we weren't logging
these events. This commit adds the missing logs.
Change-Id: I80749a7e95e8cb97bc8dd7dac1e523e223114b7f
Previously if node was not found in containment, it was
given the status, 'skipped'.
We later try to delete skipped nodes from containment.
To fix this, add a new status called 'remove' to differentiate
nodes which should be skipped and nodes which should be deleted.
Change-Id: Ic09e62dc9723c89d0c9f968ce68c039114a9d74e
We now have the piece hashes verified for all segments on all production
satellites. We can remove the code that handles the case where piece
hashes are not verified. This would make easier the migration of
services from PointerDB to the new metabase.
For consistency, PieceHashesVerified is still set to true in PointerDB
for new segments.
Change-Id: Idf0ccce4c8d01ae812f11e8384a7221d90d4c183
Make metainfo.RSConfig a valid pflag config value. This allows us to
configure the RSConfig as a string like k/m/o/n-shareSize, which makes
having multiple supported RS schemes easier in the future.
RS-related config values that are no longer needed have been removed
(MinTotalThreshold, MaxTotalThreshold, MaxBufferMem, Verify).
Change-Id: I0178ae467dcf4375c504e7202f31443d627c15e1
With the new overlay.AuditOutcome type for offline audits, the
IsUp field is redundant. If AuditOutcome != AuditOffline, then
the node is online.
In addition to removing the field itself, other changes needed
to be made regarding the relationship between 'uptime' and 'audits'.
Previously, uptime and audit outcome were completely separated. For
example, it was possible to update a node's stats to give it a
successful/failed/unknown audit while simultaneously indicating that
the node was offline by setting IsUp to false. This is no longer possible
under this changeset. Some test which did this have been changed slightly
in order to pass.
Also add new benchmarks for UpdateStats and BatchUpdateStats with different
audit outcomes.
Change-Id: I998892d615850b1f138dc62f9b050f720ea0926b
As part of the Metainfo Refactoring, we need to make the Metainfo Loop
working with both the current PointerDB and the new Metabase. Thus, the
Metainfo Loop should pass to the Observer interface more specific Object
and Segment types instead of pb.Pointer.
After this change, there are still a couple of use cases that require
access to the pb.Pointer (hence we have it as a field in the
metainfo.Segment type):
1. Expired Deletion Service
2. Repair Service
It would require additional refactoring in these two services before we
are able to clean this.
Change-Id: Ib3eb6b7507ed89d5ba745ffbb6b37524ef10ed9f
- Remove flag for switching off offline audit reporting.
- Change the overlay method used from UpdateUptime to BatchUpdateStats, as this
is where the new online scoring is done.
- Add a new overlay.AuditOutcome type: AuditOffline. Since we now use the same
method to record offline audits as success, failure, and unknown, we need to
distinguish offline audits from the rest.
Change-Id: Iadcfe10cf13466fa1a1c2dc542db8994a6423355
A year ago we made the audit service deleting expired segments.
Meanwhile, we introduced an expired deletetion sub-service in the
metainfo service which sole purpose is deleting expired segments.
Therefore, now we are removing this responsibility from the audit
service. It will continue to avoid reporting failures on expired
segments, but it would not delete them anymore.
We do this to cleanup responsibilities in advance of the metainfo
refactoring.
Change-Id: Id7aab2126f9289dbb5b0bdf7331ba7a3328730e4
This preserves the last_ip_and_port field from node lookups through
CreateAuditOrderLimits() and CreateAuditOrderLimit(), so that later
calls to (*Verifier).GetShare() can try to use that IP and port. If a
connection to the given IP and port cannot be made, or the connection
cannot be verified and secured with the target node identity, an
attempt is made to connect to the original node address instead.
A similar change is not necessary to the other Create*OrderLimits
functions, because they already replace node addresses with the cached
IP and port as appropriate. We might want to consider making a similar
change to CreateGetRepairOrderLimits(), though.
The audit situation is unique because the ramifications are especially
powerful when we get the address wrong. Failing a single audit can have
a heavy cost to a storage node. We need to make extra effort in order
to avoid imposing that cost unfairly.
Situation 1: If an audit fails because the repair worker failed to make
a DNS query (which might well be the fault on the satellite side), and
we have last_ip_and_port information available for the target node, it
would be unfair not to try connecting to that last_ip_and_port address.
Situation 2: If a node has changed addresses recently and the operator
correctly changed its DNS entry, but we don't bother querying DNS, it
would be unfair to penalize the node for our failure to connect to it.
So the audit worker must try both last_ip_and_port _and_ the node
address as supplied by the SNO.
We elect here to try last_ip_and_port first, on the grounds that (a) it
is expected to work in the large majority of cases, and (b) there
should not be any security concerns with connecting to an out-or-date
address, and (c) avoiding DNS queries on the satellite side helps
alleviate satellite operational load.
Change-Id: I9bf6c6c79866d879adecac6144a6c346f4f61200
The VerifyPieceHashes method has a sanity check for the number pieces to
be removed from the pointer after the audit for verifying the piece
hashes.
This sanity check failed when we executed the command on the production
satellites because the Verify command removes Fails and PendingAudits
nodes from the audit report if piece_hashes_verified = false.
A new temporary UsedToVerifyPieceHashes flag is added to
audits.Verifier. It is set to true only by the verify-piece-hashes
command. If the flag is true then the Verify method will always include
Fails and PendingAudits nodes in the report.
Test case is added to cover this use case.
Change-Id: I2c7cb6b12029d52b2fc565365eee0826c3de6ee8
Jira: https://storjlabs.atlassian.net/browse/PG-69
There are a number of segments with piece_hashes_verified = false in
their metadata) on US-Central-1, Europe-West-1, and Asia-East-1
satellites. Most probably, this happened due to a bug we had in the
past. We want to verify them before executing the main migration to
metabase. This would simplify the main migration to metabase with one
less issue to think about.
Change-Id: I8831af1a254c560d45bb87d7104e49abd8242236
* The audit worker wants to get items from the queue and process them.
* The audit chore wants to create new queues and swap them in when the
old queue has been processed.
This change adds a "Queues" struct which handles the concurrency
issues around the worker fetching a queue and the chore swapping a new
queue in. It simplifies the logic of the "Queue" struct to its bare
bones, so that it behaves like a normal queue with no need to understand
the details of swapping and worker/chore interactions.
Change-Id: Ic3689ede97a528e7590e98338cedddfa51794e1b
satellite.DB.Console().Projects().GetAll database query
can be replaced with planet.Uplinks[0].Projects[0].ID
Change-Id: I73b82b91afb2dde7b690917345b798f9d81f6831
If a segment is deleted, is modified, or expires during an audit, this
is not problematic, so we should not return errors. Functionally,
nothing changes, but our metrics around audit success rate will be
improved after this change.
Change-Id: Ic11df056b2c73894b67a55894bd4d58c00470606
* Do not swap the active audit queue with the pending audit queue until
the active audit queue is empty.
* Do not begin creating a new pending audit queue until the existing
pending audit queue has been swapped to the active queue.
Change-Id: I81db5bfa01458edb8cdbe71f5baeebdcb1b94317
Since we increased the number of concurrent audit workers to two, there are going
to be instances of a single node being audited simultaneously for different segments.
If the node times out for both, we will try to write them both to the pending audits
table, and the second will return an error since the path is not the same as what
already exists. Since with concurrent workers this is expected, we will log the
occurrence rather than return an error.
Since the release default audit concurrency is 2, update testplanet default to run with
concurrent workers as well.
Change-Id: I4e657693fa3e825713a219af3835ae287bb062cb
This test failed due to a timeout on a download which is supposed to
succeed. The testplanet default for the value is 5 seconds, but here
it is 500 milliseconds.
It looks like this is due to the fact that later in the test we need to
wait for a slow node to timeout, so we cut the timeout shorter to reduce
test time.
This PR increases the timeout to 1 second. Still not too long to wait, but
gives us twice as much time to download, decreasing the likelihood that we
see the timeout error.
Change-Id: I504db39ab5dc4d3c505520337b258265d6da7020
What: As soon as a node passes the vetting criteria (total_audit_count and total_uptime_count
are greater than the configured thresholds), we set vetted_at to the current timestamp.
Why: We may want to use this timestamp in future development to select new vs vetted nodes.
It also allows flexibility in node vetting experiments and allows for better metrics around
vetting times.
Please describe the tests: satellitedb_test: TestUpdateStats and TestBatchUpdateStats make sure vetted_at is set appropriately
Please describe the performance impact: This change does add extra logic to BatchUpdateStats and UpdateStats and
commits another variable to the db (vetted_at), but this should be negligible.
Change-Id: I3de804549b5f1bc359da4935bc859758ceac261d
Also distinguish the purpose for selecting nodes to avoid potential
confusion, what should allow caching and what shouldn't.
Change-Id: Iee2451c1f10d0f1c81feb1641507400d89918d61
Add a flag that allows us to easily switch disqualification from
suspension mode on or off. A node will only be disqualified from
suspension mode if it has been suspended for longer than the grace
period AND the SuspensionDQEnabled flag is true.
Change-Id: I9e67caa727183cd52ab2042b0a370a1bcaebe792
TestVerifierSlowDownload would sometimes not have enough nodes finish in
the allotted deadline period. This increases the deadline and also does
not assert that exactly 3 have finished. Instead, in keeping with the
purpose of the test, it asserts that the slow download is never counted
as a success and is always counted as a pending audit in the final
report.
Change-Id: I180734fcc4a499420c75164bad6253ed155d87de
Sometimes nodes who have gracefully exited will still be holding pieces
according to the satellite. This has some unintended side effects
currently, such as nodes getting disqualified after having successfully
exited.
* When the audit reporter attempts to update node stats, do not update
stats (alpha, beta, suspension, disqualification) if the node has
finished graceful exit (audit/reporter_test.go TestGracefullyExitedNotUpdated)
* Treat gracefully exited nodes as "not reputable" so that the repairer
and checker do not count them as healthy (overlay/statdb_test.go
TestKnownUnreliableOrOffline, repair/repair_test.go
TestRepairGracefullyExited)
Change-Id: I1920d60dd35de5b2385a9b06989397628a2f1272
If a node is suspended and receives an unknown or failing audit,
disqualify them if the grace period (default 1w in production) has
passed.
Migrate the nodes table so any node that is currently suspended gets
unsuspended when the satellite starts up.
Change-Id: I7b81c68026f823417faa0bf5e5cb5e67c7156b82
Alpha=1 and beta=0 are the expected first values for any alpha/beta
reputation system we are using in the codebase. So we are removing the
configurability of these values.
Change-Id: Ic61861b8ea5047fa1438ea6609b1d0048bf0abc3
Whenever the node's reputation is updated, if its unknown audit
reputation is below the suspension threshold, its suspension field
is set to the current time. This could overwrite the previous
"suspendedAt" value resulting a node that never reaches the end of
its suspension.
Also log whenever a node is disqualified or its suspension status
changes
Change-Id: I5e8c8f1c46f66d79cb279b5b16a84fe03f533deb
We have been using the SQL expression `name='(*Verifier).Verify' AND
error_name='not enough shares for successful audit'` thus far to detect
cases of this problem and alert on them. Unfortunately, since this
rarely (hopefully never) happens, influxdb has no data for most of the
auditor instances, and when it has no data for a time series, it returns
no columns either. This makes Redash upset when it tries to perform a
query for an alert and can't find the column whose value it expects to
check.
This change should make it so zero values are reported when the problem
has not happened, and higher values when it has.
Change-Id: I79e5e000f879678b661dac88caae1e2915b39ab1
Instead of providing the database from outside to testplanet create it
inside and then allow wrapping and modifying it. This is more convenient
to use.
Change-Id: I9b8f69e6e0a19ff984b4e2bfe927c9100c77bc6c
Make sure that suspended nodes are treated appropriately by the overlay
cache. This means we should expect the following behavior:
* suspended nodes (vetted or not) should not be selected for uploading
new segments
* suspended nodes should be treated by the checker and repairer as
"unhealthy", and should be removed upon successful repair
This commit also removes unused overlay functionality.
Fixes a bug with commit 8b72181a1f where
the audit reporter was automatically suspending nodes regardless of
audit outcome (see test added).
Tests:
* updates repair tests to ensure that a suspended node is treated as
unhealthy and will be removed from the pointer on successful repair
* updates overlay tests for KnownUnreliableOrOffline and KnownReliable
to expect suspended nodes to be considered "unreliable"
* adds satellitedb test that ensures overlay.SelectStorageNodes and
overlay.SelectNewStorageNodes do not include suspended nodes
* adds audit reporter test to ensure that different audit outcomes
result in the correct suspended/disqualified states
Change-Id: I40dba67278c8e8d2ce0bcec5e0a5cb6e4ce2f561
* change overlay.UpdateStats to allow a third audit outcome. Now it can
handle successful, failed, and unknown audits.
* when "unknown audit reputation"
(unknownAuditAlpha/(unknownAuditAlpha+unknownAuditBeta)) falls below the
DQ threshold, put node into suspension.
* when unknown audit reputation goes above the DQ threshold, remove node
from suspension.
* record unknown audits from audit reporter.
* add basic tests around unknown audits and suspension.
Change-Id: I125f06f3af52e8a29ba48dc19361821a9ff1daa1
- Previously, checkSegmentAltered only checked for segments that were replaced
but we want to detect all changes to a segment that occurred while an audit was being conducted.
- Fixed a bug where nodes failing audits during reverify for non-piece-hash-verified
segments were not being removed from containment mode.
- Filled in gaps in reverify testing to ensure nodes are properly removed from containment.
Change-Id: Icd96d369278987200fd28581395725438972b292
On satellite, remove all references to free_bandwidth column in nodes table.
On storage node, remove references to AllocatedBandwidth and MinimumBandwidth and mark as deprecated.
Protobuf message, NodeCapacity, is left intact for backwards compatibility.
Once this is released to all satellites, we can drop the column from the DB.
Change-Id: I2ff6c6537fc9008a0c5588e951afea58ede85838
Currently SNs report their free disk space once per hour. If a node
becomes full, it has to wait until the next contact cycle begins to
report; all the while receiving and failing upload requests. By increasing
the minimum required disk space, we can give the storage nodes more time
to report their space before the completely fill up. This change goes
hand-in-hand with another change we want to implement: trigger capacity
report on SN immediately upon falling below threshold.
Change-Id: I12f778286c6c3f582438b0e2949765ac43325e27
this commit updates our monkit dependency to the v3 version where
it outputs in an influx style. this makes discovery much easier
as many tools are built to look at it this way.
graphite and rothko will suffer some due to no longer being a tree
based on dots. hopefully time will exist to update rothko to
index based on the new metric format.
it adds an influx output for the statreceiver so that we can
write to influxdb v1 or v2 directly.
Change-Id: Iae9f9494a6d29cfbd1f932a5e71a891b490415ff
With this change RS configuration will be set on satellite. Uplink with
get RS values with BeginObject request and will use it. For backward
compatibility and to avoid super large change redundancy scheme stored
with bucket is not touched. This can be done in future.
Change-Id: Ia5f76fc10c37e2c44e4f7b8754f28eafe1f97eff
With the new storage node downtime tracking feature, we need remove current uptime reputation configs: UptimeReputationAlpha, UptimeReputationBeta, and
UptimeReputationDQ. This is the first step of removing the uptime
reputation columns from satellitedb
Change-Id: Ie8fab13295dbf545e33aeda0c4306cda4ba54e36
Remove starting up messages from peers. We expect all of them to start,
if they don't, then they should return an error why they don't start.
The only informative message is when a service is disabled.
When doing initial database setup then each migration step isn't
informative, hence print only a single line with the final version.
Also use shorter log scopes.
Change-Id: Ic8b61411df2eeae2a36d600a0c2fbc97a84a5b93
Remove direct dependency on uplink.RSConfig, this simplifies
moving the config file without introducing weird dependencies.
Change-Id: I7fd2a145401e0205d7047631df9d2810241efeec
