* feat: generate setting plan and simplicy the logic and cache the plan for reuse
Signed-off-by: Harshit Gangal <harshit@planetscale.com>
* feat: reset settings over closing when settings do not match
Signed-off-by: Harshit Gangal <harshit@planetscale.com>
* refactor: setting interface to struct
Signed-off-by: Harshit Gangal <harshit@planetscale.com>
* test: inifinite session test having connection settings
Signed-off-by: Harshit Gangal <harshit@planetscale.com>
Signed-off-by: Harshit Gangal <harshit@planetscale.com>
* decouple olap tx timeout from oltp tx timeout
Since workload=olap bypasses the query timeouts
(--queryserver-config-query-timeout) and also row limits, the natural
assumption is that it also bypasses the transaction timeout.
This is not the case, e.g. for a tablet where the
--queryserver-config-transaction-timeout is 10.
This commit:
* Adds new CLI flag and YAML field to independently configure TX
timeouts for OLAP workloads (--queryserver-config-olap-transaction-timeout).
* Decouples TX kill interval from OLTP TX timeout via new CLI flag and
YAML field (--queryserver-config-transaction-killer-interval).
Signed-off-by: Max Englander <max@planetscale.com>
* decouple ol{a,t}p tx timeouts: pr comments #1
Signed-off-by: Max Englander <max@planetscale.com>
* decouple ol{a,t}p tx timeouts: pr comments #2 consolidate timeout logic in sc
Signed-off-by: Max Englander <max@planetscale.com>
* decouple ol{a,t}p tx timeouts: remove unused tx killer flag
Signed-off-by: Max Englander <max@planetscale.com>
* decouple ol{a,t}p tx timeouts: update 15_0_0_summary.md
Signed-off-by: Max Englander <max@planetscale.com>
* decouple ol{a,t}p tx timeouts: fix race cond
Signed-off-by: Max Englander <max@planetscale.com>
* decouple ol{a,t}p tx timeouts: pr comments #3 -txProps.timeout, +sc.expiryTime
Signed-off-by: Max Englander <max@planetscale.com>
* decouple ol{a,t}p tx timeouts: pr comments #4 -atomic.Value for expiryTime
Signed-off-by: Max Englander <max@planetscale.com>
* decouple ol{a,t}p tx timeouts: fix race cond (without atomic.Value)
Signed-off-by: Max Englander <max@planetscale.com>
* decouple ol{a,t}p tx timeouts: pr comments #5 -unused funcs, fix comments, set ticks interval once
Signed-off-by: Max Englander <max@planetscale.com>
* decouple ol{a,t}p tx timeouts: pr comments #5 +txkill tests
Signed-off-by: Max Englander <max@planetscale.com>
* revert fmt changes
Signed-off-by: Max Englander <max@planetscale.com>
* implement pr review suggestion
Signed-off-by: Max Englander <max@planetscale.com>
Signed-off-by: Max Englander <max@planetscale.com>
* feat: added connections with settings logic to resource pool
Signed-off-by: Harshit Gangal <harshit@planetscale.com>
* feat: support refresh pool for settings connection channel and added benchmark for settings connection
Signed-off-by: Harshit Gangal <harshit@planetscale.com>
* feat: take from settings channel and reset the conn if nothing present in non-setting channel
Signed-off-by: Harshit Gangal <harshit@planetscale.com>
* fix: return 0 for setting hash for resource where it does not exists
Signed-off-by: Harshit Gangal <harshit@planetscale.com>
* test: added benchmark for applying settings individual set statement and single statement
Signed-off-by: Harshit Gangal <harshit@planetscale.com>
* test: added mixed settings benchmarks
Signed-off-by: Harshit Gangal <harshit@planetscale.com>
* feat: close idle resources in settings pool andadded unit tests
Signed-off-by: Harshit Gangal <harshit@planetscale.com>
* refactor: removed settingHash from Put method of pool
Signed-off-by: Harshit Gangal <harshit@planetscale.com>
* fix: better channel select in get, avoid closing connection when no setting is applied on the connection in getwithsettings and only apply the settings on that connection
Signed-off-by: Harshit Gangal <harshit@planetscale.com>
* test: added addtional unit test
Signed-off-by: Harshit Gangal <harshit@planetscale.com>
* feat: compare the two settings rather than using hash
Signed-off-by: Harshit Gangal <harshit@planetscale.com>
* added comment for a SetCapacity code for better explaination
Signed-off-by: Harshit Gangal <harshit@planetscale.com>
* select channel is enough to block till a resource is not received, removed for loop around it
Signed-off-by: Harshit Gangal <harshit@planetscale.com>
Signed-off-by: Harshit Gangal <harshit@planetscale.com>
* Run latest gofmt on everything & address linter warnings
Signed-off-by: Matt Lord <mattalord@gmail.com>
* Rebuild proto impls with new protoc
Signed-off-by: Matt Lord <mattalord@gmail.com>
* Field name is needed for zk
Signed-off-by: Matt Lord <mattalord@gmail.com>
* Make protos with 1.18.4
Signed-off-by: Matt Lord <mattalord@gmail.com>
Signed-off-by: Matt Lord <mattalord@gmail.com>
* refactor: add resource pool as interface
Signed-off-by: Harshit Gangal <harshit@planetscale.com>
* test: mark resource pool test as non-flaky
Signed-off-by: Harshit Gangal <harshit@planetscale.com>
* test: adds a benchmark test
Signed-off-by: Harshit Gangal <harshit@planetscale.com>
* refactor: pull refresh pool code out from pool implementation
Signed-off-by: Harshit Gangal <harshit@planetscale.com>
* addressed review comments
Signed-off-by: Harshit Gangal <harshit@planetscale.com>
Previously, the transaction pool overrode the error codes for pool timeouts expiring. Let's use consistent error codes across the pools.
Signed-off-by: dleibovic <dleibovic@etsy.com>
general. We now drain the pool and re-open when we see a DNS change,
instead of reconnecting when the connection is pulled from the pool.
Signed-off-by: Jacques Grove <aquarapid@gmail.com>
An RPCPool is a special case of the ResourcePool, where the actual
"resources" in the pool are essentially a lease to perform an RPC (or
access any shared resource) with differing timeout semantics than the
sync2.Semaphore.
Signed-off-by: Andrew Mason <amason@slack-corp.com>
The `cache.LRUCache` struct has now been abstracted behind a generic
Cache interface so that it can be swapped with more complex Cache
implementations.
Signed-off-by: Vicent Marti <vmg@strn.cat>
* Rename sc to bc; legacy of SlaveConnection
* Rename connID to serverID to reflect what it actually represents
Signed-off-by: Jacques Grove <aquarapid@gmail.com>
The DBConnPool stats vars were not used by anyone other than
vttablet. So, I've moved them into tabletserver's connnpool,
which simplifies DBConnPool.
Signed-off-by: Sugu Sougoumarane <ssougou@gmail.com>
* This PR adds a new error that distinguishes between actual timeouts from the
pool and errors due to contexts being already expired by the time it gets to
the pool.
Signed-off-by: Rafael Chacon <rafael@slack-corp.com>
Background: this change allows one to create prefilled resource
pools. This is useful when traffics suddenly shift while the
pool is still empty. This causes a thundering herd of Open requests
that can cause outage.
There is a proposal to rewrite this pool to natively accommodate
this feature. This change is a stand-in until that effort is
completed.
Signed-off-by: Sugu Sougoumarane <ssougou@gmail.com>
Doing object allocation and fmt.Sprintf under a lock is unecessary and
expensive. Refactor Register/Unregister to reduce blocking.
Benchmark shows A change. Probably inconsequential, but can't hurt.
Signed-off-by: Daniel Tahara <tahara@dropbox.com>
This commit consists only of the changes from running the following:
```
find go -name '*.go' | xargs -n 10 -- sed -i -r -e \
's,"github.com/youtube/vitess/go/,"vitess.io/vitess/go/,'
```
Harshit Gangal
Max Englander
Matt Lord
Vicent Marti
dleibovic
Jacques Grove
Andres Taylor
Andrew Mason
Aditya Sharma
imxyb
Sugu Sougoumarane
Rafael Chacon
Lloyd Cabancla
Morgan Tocker
yuxiaobo
Adam Saponara
Enrique Bris
Alec Thomas
Danny Wensley
deepthi
Daniel Tahara
Anthony Yeh