Skip to content

Commit 823194d

Browse files
andythsumosabua
authored andcommitted
Update operation.md
1 parent 80854d3 commit 823194d

File tree

4 files changed

+18
-15
lines changed

4 files changed

+18
-15
lines changed
35.5 KB
Loading
-34.4 KB
Binary file not shown.
-4.26 KB
Loading

docs/operation.md

Lines changed: 18 additions & 15 deletions
Original file line numberDiff line numberDiff line change
@@ -9,12 +9,12 @@ Trino Gateway records history of recent queries and displays links to check quer
99
details page in respective trino cluster.
1010
![trino.gateway.io](./assets/trinogateway_query_history.png)
1111

12-
## Admin UI - add and modify backend information
12+
## Cluster UI - add and modify cluster information
1313

14-
The admin page is used to configure the gateway to multiple backends.
15-
Existing backend information can also be modified using the same.
14+
The cluster page is used to configure Trino Gateway for multiple Trino clusters.
15+
Existing cluster information can also be modified using the edit button.
1616

17-
![trino.gateway.io/entity](./assets/trinogateway_ha_admin.png)
17+
![trino.gateway.io/entity](./assets/trinogateway_cluster_page.png)
1818

1919

2020
## Graceful shutdown
@@ -25,29 +25,32 @@ Query ID.
2525

2626
To graceful shutdown a trino cluster without query losses, the steps are:
2727

28-
1. Set the backend to deactivate state, this prevents any new incoming queries
29-
from getting assigned to the backend.
30-
2. Poll the trino backend coorinator URL until the queued query count and the
31-
running query count both hit 0.
32-
3. Terminate the trino Coordinator & Worker Java process.
28+
1. Deactivate the cluster by turning off the 'Active' switch. This ensures that no
29+
new incoming queries are routed to the cluster.
30+
2. Poll the Trino cluster coordinator URL until the queued query count and the
31+
running query count are both zero.
32+
3. Terminate the Trino coordinator and worker Java processes.
3333

3434
To gracefully shutdown a single worker process, refer to the [Trino
3535
documentation](https://trino.io/docs/current/admin/graceful-shutdown.html) for
3636
more details.
3737

3838
## Query routing options
3939

40-
- The default router selects the backend randomly to route the queries.
41-
- If you want to route the queries to the least loaded backend for a user
42-
i.e. backend with the least number of queries running or queued from a particular user,
43-
then use `QueryCountBasedRouter`, it can be configured by adding the module name
44-
to config file's modules section like below
40+
- The default router selects the cluster randomly to route the queries.
41+
- If you want to route the queries to the least loaded cluster for a user
42+
so the cluster with the fewest running or queued queries,
43+
use `QueryCountBasedRouter`. You can enable it by adding the module name
44+
to the `modules` section of the config file:
4545

4646
```yaml
4747
modules:
4848
- io.trino.gateway.ha.module.QueryCountBasedRouterProvider
4949
```
50-
- The router works on the stats it receives from the clusters about the load i.e number queries queued and running on a cluster at regular intervals which can be configured like below. The default interval is 1 min
50+
- The router operates based on the stats it receives from the clusters, such as
51+
the number of queued and running queries. These values are retrieved at regular
52+
intervals. This interval can be configured by setting `taskDelaySeconds` under
53+
`monitor` section in the config file. The default interval is 60
5154
```yaml
5255
monitor:
5356
taskDelaySeconds: 10

0 commit comments

Comments
 (0)