From 53d41619365b5629d749e95f1831d8848e0f458a Mon Sep 17 00:00:00 2001 From: Luke Butters Date: Thu, 25 Jun 2026 16:14:28 +1000 Subject: [PATCH 01/10] Add polling connection count troubleshooting section Document how to resolve polling Tentacle request timeouts that occur while the Tentacle is online and busy, by increasing the polling connection count. Co-Authored-By: Claude Opus 4.8 (1M context) --- .../troubleshooting-polling.mdx | 30 +++++++++++++++++++ 1 file changed, 30 insertions(+) diff --git a/src/pages/docs/infrastructure/deployment-targets/tentacle/troubleshooting/troubleshooting-polling.mdx b/src/pages/docs/infrastructure/deployment-targets/tentacle/troubleshooting/troubleshooting-polling.mdx index f4ddc3e49d..0ee8cd2daf 100644 --- a/src/pages/docs/infrastructure/deployment-targets/tentacle/troubleshooting/troubleshooting-polling.mdx +++ b/src/pages/docs/infrastructure/deployment-targets/tentacle/troubleshooting/troubleshooting-polling.mdx @@ -106,4 +106,34 @@ To do this open **Control Panel ➜ Internet Options ➜ Advanced**, and uncheck ![](/docs/img/infrastructure/deployment-targets/tentacle/images/5865771.png) ::: +## Requests timing out while the Tentacle is online + +If a health check or deployment fails with an error like: + +> A request was sent to a polling endpoint, but the polling endpoint did not collect the request within the allowed time (00:02:00), so the request timed out. + +and the Tentacle is online and *is* successfully processing other requests, the Tentacle may simply be unable to collect requests fast enough to keep up. + +A polling Tentacle can only work on as many requests at once as it has open polling connections to the Octopus Server, and each connection handles a single request at a time. Historically polling Tentacles were configured with a connection count of `1`, which means a single long-running request — for example transferring a large package — blocks every other request behind it. Those queued requests can eventually time out with the error above even though the Tentacle itself is healthy and busy. + +You can resolve this by increasing the number of polling connections the Tentacle opens, allowing it to process more requests concurrently. + +Set it from the Tentacle command line: + +```bash +Tentacle set-polling-connection-count --instance=MyInstance --pollingConnectionCount=16 +``` + +Or set it directly in the Tentacle configuration file: + +```xml +16 +``` + +Restart the Tentacle for the change to take effect. The value is clamped to between `1` and `512`. + +:::div{.hint} +New polling Tentacle workers are automatically configured with a connection count of `max(5, processor count)`. Deployment targets — and any Tentacle configured before this option existed — default to `1`, so they are the most likely to benefit from increasing the count. +::: + From 2d807beb35ad4c0f4dcd59d39fc9d6438620d79b Mon Sep 17 00:00:00 2001 From: Luke Butters Date: Thu, 25 Jun 2026 16:25:38 +1000 Subject: [PATCH 02/10] Add alt text to Octopus Server configured successfully image Co-Authored-By: Claude Opus 4.8 (1M context) --- .../tentacle/troubleshooting/troubleshooting-polling.mdx | 2 +- 1 file changed, 1 insertion(+), 1 deletion(-) diff --git a/src/pages/docs/infrastructure/deployment-targets/tentacle/troubleshooting/troubleshooting-polling.mdx b/src/pages/docs/infrastructure/deployment-targets/tentacle/troubleshooting/troubleshooting-polling.mdx index 0ee8cd2daf..e65e11706d 100644 --- a/src/pages/docs/infrastructure/deployment-targets/tentacle/troubleshooting/troubleshooting-polling.mdx +++ b/src/pages/docs/infrastructure/deployment-targets/tentacle/troubleshooting/troubleshooting-polling.mdx @@ -37,7 +37,7 @@ Octopus Cloud instances have HSTS enabled so it will be impossible to bypass the The page shown should look like the one below. :::figure -![](/docs/img/infrastructure/deployment-targets/tentacle/images/3277906.png) +![Octopus Server configured successfully page](/docs/img/infrastructure/deployment-targets/tentacle/images/3277906.png) ::: If you've made it this far, good news! Your Octopus Server is running and ready to accept inbound connections from Polling Tentacles. From 64fc56e28875aade6ba8f19daf02795c77977189 Mon Sep 17 00:00:00 2001 From: Luke Butters Date: Thu, 25 Jun 2026 16:26:52 +1000 Subject: [PATCH 03/10] Add alt text to certificate revocation control panel image Co-Authored-By: Claude Opus 4.8 (1M context) --- .../tentacle/troubleshooting/troubleshooting-polling.mdx | 2 +- 1 file changed, 1 insertion(+), 1 deletion(-) diff --git a/src/pages/docs/infrastructure/deployment-targets/tentacle/troubleshooting/troubleshooting-polling.mdx b/src/pages/docs/infrastructure/deployment-targets/tentacle/troubleshooting/troubleshooting-polling.mdx index e65e11706d..7a3cd8af12 100644 --- a/src/pages/docs/infrastructure/deployment-targets/tentacle/troubleshooting/troubleshooting-polling.mdx +++ b/src/pages/docs/infrastructure/deployment-targets/tentacle/troubleshooting/troubleshooting-polling.mdx @@ -103,7 +103,7 @@ If the command help is not displayed immediately (< 1s) you may need to consider To do this open **Control Panel ➜ Internet Options ➜ Advanced**, and uncheck the *Check for publisher's certificate revocation* option as shown below. :::figure -![](/docs/img/infrastructure/deployment-targets/tentacle/images/5865771.png) +![uncheck of the Check for publisher’s certificate revocation control panel option](/docs/img/infrastructure/deployment-targets/tentacle/images/5865771.png) ::: ## Requests timing out while the Tentacle is online From de0a9146f36fcc20124390f2efa47d32becee174 Mon Sep 17 00:00:00 2001 From: Luke Butters Date: Thu, 25 Jun 2026 16:36:00 +1000 Subject: [PATCH 04/10] Reword polling connection count timeout troubleshooting section Co-Authored-By: Claude Opus 4.8 (1M context) --- .../troubleshooting-polling.mdx | 21 +++++++++++-------- 1 file changed, 12 insertions(+), 9 deletions(-) diff --git a/src/pages/docs/infrastructure/deployment-targets/tentacle/troubleshooting/troubleshooting-polling.mdx b/src/pages/docs/infrastructure/deployment-targets/tentacle/troubleshooting/troubleshooting-polling.mdx index 7a3cd8af12..92a5c34427 100644 --- a/src/pages/docs/infrastructure/deployment-targets/tentacle/troubleshooting/troubleshooting-polling.mdx +++ b/src/pages/docs/infrastructure/deployment-targets/tentacle/troubleshooting/troubleshooting-polling.mdx @@ -106,34 +106,37 @@ To do this open **Control Panel ➜ Internet Options ➜ Advanced**, and uncheck ![uncheck of the Check for publisher’s certificate revocation control panel option](/docs/img/infrastructure/deployment-targets/tentacle/images/5865771.png) ::: -## Requests timing out while the Tentacle is online +## Requests timing out during Deployments/Runbook while the Tentacle is online -If a health check or deployment fails with an error like: +If the tentacle is online and if while running multiple Deployments and Runbooks the following error: > A request was sent to a polling endpoint, but the polling endpoint did not collect the request within the allowed time (00:02:00), so the request timed out. -and the Tentacle is online and *is* successfully processing other requests, the Tentacle may simply be unable to collect requests fast enough to keep up. +is occasionally seen. It could be that the Polling Tentacle is unable to process the requests quickly enough, despite the Tentacle's CPU and other +resources being under utilised. -A polling Tentacle can only work on as many requests at once as it has open polling connections to the Octopus Server, and each connection handles a single request at a time. Historically polling Tentacles were configured with a connection count of `1`, which means a single long-running request — for example transferring a large package — blocks every other request behind it. Those queued requests can eventually time out with the error above even though the Tentacle itself is healthy and busy. +This can occure because Polling Tentacle may be configured to process no more than one requests at a time, as their polling connect count is set to `1` or +defaults to `1`. Polling Tentacle Targets are by default configured to have a polling connection count of 1. While Polling Workers default to `max(5, Tentacle Core Count)`, +however historically theu have been configured with a Polling Connection Count of `1`. -You can resolve this by increasing the number of polling connections the Tentacle opens, allowing it to process more requests concurrently. +To overcome this the polling connection count can be increased. For workers the recommended value is `max(5, Tentacle Core Count)`. Set it from the Tentacle command line: ```bash -Tentacle set-polling-connection-count --instance=MyInstance --pollingConnectionCount=16 +Tentacle set-polling-connection-count --instance=MyInstance --pollingConnectionCount=5 ``` Or set it directly in the Tentacle configuration file: ```xml -16 +5 ``` -Restart the Tentacle for the change to take effect. The value is clamped to between `1` and `512`. +Restart the Tentacle for the change to take effect. :::div{.hint} -New polling Tentacle workers are automatically configured with a connection count of `max(5, processor count)`. Deployment targets — and any Tentacle configured before this option existed — default to `1`, so they are the most likely to benefit from increasing the count. +New Polling Tentacle workers are automatically configured with a connection count of `max(5, processor count)`. Deployment targets — and any Tentacle configured before this option existed — default to `1`, so they are the most likely to benefit from increasing the count. ::: From c17f551ef8b3bf6253c92cb825e53ae651034b8a Mon Sep 17 00:00:00 2001 From: Luke Butters Date: Thu, 25 Jun 2026 16:38:14 +1000 Subject: [PATCH 05/10] Fix typos in polling connection count troubleshooting section Co-Authored-By: Claude Opus 4.8 (1M context) --- .../tentacle/troubleshooting/troubleshooting-polling.mdx | 6 +++--- 1 file changed, 3 insertions(+), 3 deletions(-) diff --git a/src/pages/docs/infrastructure/deployment-targets/tentacle/troubleshooting/troubleshooting-polling.mdx b/src/pages/docs/infrastructure/deployment-targets/tentacle/troubleshooting/troubleshooting-polling.mdx index 92a5c34427..254a2772d1 100644 --- a/src/pages/docs/infrastructure/deployment-targets/tentacle/troubleshooting/troubleshooting-polling.mdx +++ b/src/pages/docs/infrastructure/deployment-targets/tentacle/troubleshooting/troubleshooting-polling.mdx @@ -106,7 +106,7 @@ To do this open **Control Panel ➜ Internet Options ➜ Advanced**, and uncheck ![uncheck of the Check for publisher’s certificate revocation control panel option](/docs/img/infrastructure/deployment-targets/tentacle/images/5865771.png) ::: -## Requests timing out during Deployments/Runbook while the Tentacle is online +## Requests timing out during Deployments/Runbooks while the Tentacle is online If the tentacle is online and if while running multiple Deployments and Runbooks the following error: @@ -115,9 +115,9 @@ If the tentacle is online and if while running multiple Deployments and Runbooks is occasionally seen. It could be that the Polling Tentacle is unable to process the requests quickly enough, despite the Tentacle's CPU and other resources being under utilised. -This can occure because Polling Tentacle may be configured to process no more than one requests at a time, as their polling connect count is set to `1` or +This can occur because a Polling Tentacle may be configured to process no more than one request at a time, as their polling connection count is set to `1` or defaults to `1`. Polling Tentacle Targets are by default configured to have a polling connection count of 1. While Polling Workers default to `max(5, Tentacle Core Count)`, -however historically theu have been configured with a Polling Connection Count of `1`. +historically they have been configured with a Polling Connection Count of `1`. To overcome this the polling connection count can be increased. For workers the recommended value is `max(5, Tentacle Core Count)`. From d1340b386bb92bdd75b5300d2e272754c9a3d22f Mon Sep 17 00:00:00 2001 From: Luke Butters Date: Thu, 25 Jun 2026 16:41:52 +1000 Subject: [PATCH 06/10] Clarify wording in polling connection count troubleshooting section Co-Authored-By: Claude Opus 4.8 (1M context) --- .../tentacle/troubleshooting/troubleshooting-polling.mdx | 2 +- 1 file changed, 1 insertion(+), 1 deletion(-) diff --git a/src/pages/docs/infrastructure/deployment-targets/tentacle/troubleshooting/troubleshooting-polling.mdx b/src/pages/docs/infrastructure/deployment-targets/tentacle/troubleshooting/troubleshooting-polling.mdx index 254a2772d1..c55096f39e 100644 --- a/src/pages/docs/infrastructure/deployment-targets/tentacle/troubleshooting/troubleshooting-polling.mdx +++ b/src/pages/docs/infrastructure/deployment-targets/tentacle/troubleshooting/troubleshooting-polling.mdx @@ -108,7 +108,7 @@ To do this open **Control Panel ➜ Internet Options ➜ Advanced**, and uncheck ## Requests timing out during Deployments/Runbooks while the Tentacle is online -If the tentacle is online and if while running multiple Deployments and Runbooks the following error: +If the Polling Tentacle is online and if while running multiple Deployments or Runbooks the following error: > A request was sent to a polling endpoint, but the polling endpoint did not collect the request within the allowed time (00:02:00), so the request timed out. From 037cfff68d4ba068f3fb58222dfc0bdf6311f6f1 Mon Sep 17 00:00:00 2001 From: Luke Butters Date: Thu, 25 Jun 2026 16:46:35 +1000 Subject: [PATCH 07/10] Fix MD036 lint and backtick consistency in polling troubleshooting section Add colons to bold labels so markdownlint MD036 doesn't flag them as headings, and wrap max(5, Tentacle Core Count) in backticks consistently. Co-Authored-By: Claude Opus 4.8 (1M context) --- .../troubleshooting-polling.mdx | 21 +++++++++---------- 1 file changed, 10 insertions(+), 11 deletions(-) diff --git a/src/pages/docs/infrastructure/deployment-targets/tentacle/troubleshooting/troubleshooting-polling.mdx b/src/pages/docs/infrastructure/deployment-targets/tentacle/troubleshooting/troubleshooting-polling.mdx index c55096f39e..eeba89d98d 100644 --- a/src/pages/docs/infrastructure/deployment-targets/tentacle/troubleshooting/troubleshooting-polling.mdx +++ b/src/pages/docs/infrastructure/deployment-targets/tentacle/troubleshooting/troubleshooting-polling.mdx @@ -108,18 +108,21 @@ To do this open **Control Panel ➜ Internet Options ➜ Advanced**, and uncheck ## Requests timing out during Deployments/Runbooks while the Tentacle is online -If the Polling Tentacle is online and if while running multiple Deployments or Runbooks the following error: +If a Polling Tentacle is online but you occasionally see timeout errors during concurrent deployments or runbooks, the Tentacle may be hitting its connection limit. + +**The Error:** > A request was sent to a polling endpoint, but the polling endpoint did not collect the request within the allowed time (00:02:00), so the request timed out. -is occasionally seen. It could be that the Polling Tentacle is unable to process the requests quickly enough, despite the Tentacle's CPU and other -resources being under utilised. +**The Cause:** + +This issue occurs when a Polling Tentacle cannot process requests quickly enough, even if the host CPU and RAM utilization are low. By default, Polling Tentacle Deployment Targets are configured with a polling connection count of 1. + +While new Polling Workers automatically scale to `max(5, Tentacle Core Count)`, older Tentacles and standard Deployment Targets default to 1 and only process one request at a time. -This can occur because a Polling Tentacle may be configured to process no more than one request at a time, as their polling connection count is set to `1` or -defaults to `1`. Polling Tentacle Targets are by default configured to have a polling connection count of 1. While Polling Workers default to `max(5, Tentacle Core Count)`, -historically they have been configured with a Polling Connection Count of `1`. +**The Solution:** -To overcome this the polling connection count can be increased. For workers the recommended value is `max(5, Tentacle Core Count)`. +Increase the polling connection count to allow concurrent request processing. The recommended value for targets and workers is `max(5, Tentacle Core Count)`. Set it from the Tentacle command line: @@ -135,8 +138,4 @@ Or set it directly in the Tentacle configuration file: Restart the Tentacle for the change to take effect. -:::div{.hint} -New Polling Tentacle workers are automatically configured with a connection count of `max(5, processor count)`. Deployment targets — and any Tentacle configured before this option existed — default to `1`, so they are the most likely to benefit from increasing the count. -::: - From a53641fcd74d28d837833bc2d61f0e013625392f Mon Sep 17 00:00:00 2001 From: Luke Butters Date: Thu, 25 Jun 2026 16:49:17 +1000 Subject: [PATCH 08/10] Refine polling connection count troubleshooting wording Co-Authored-By: Claude Opus 4.8 (1M context) --- .../tentacle/troubleshooting/troubleshooting-polling.mdx | 4 ++-- 1 file changed, 2 insertions(+), 2 deletions(-) diff --git a/src/pages/docs/infrastructure/deployment-targets/tentacle/troubleshooting/troubleshooting-polling.mdx b/src/pages/docs/infrastructure/deployment-targets/tentacle/troubleshooting/troubleshooting-polling.mdx index eeba89d98d..f57414d77e 100644 --- a/src/pages/docs/infrastructure/deployment-targets/tentacle/troubleshooting/troubleshooting-polling.mdx +++ b/src/pages/docs/infrastructure/deployment-targets/tentacle/troubleshooting/troubleshooting-polling.mdx @@ -116,9 +116,9 @@ If a Polling Tentacle is online but you occasionally see timeout errors during c **The Cause:** -This issue occurs when a Polling Tentacle cannot process requests quickly enough, even if the host CPU and RAM utilization are low. By default, Polling Tentacle Deployment Targets are configured with a polling connection count of 1. +This issue occurs when a Polling Tentacle cannot process requests quickly enough, even if the host CPU and RAM utilization are low. By default, Polling Tentacle Deployment Targets are configured with a polling connection count of `1`. -While new Polling Workers automatically scale to `max(5, Tentacle Core Count)`, older Tentacles and standard Deployment Targets default to 1 and only process one request at a time. +While newly configured Polling Workers automatically scale to `max(5, Tentacle Core Count)`, previously configured Tentacles and standard Deployment Targets default to `1` and only process one request at a time. **The Solution:** From ec48f9f5cdcae10f38e64f329fe428912300acf7 Mon Sep 17 00:00:00 2001 From: Luke Butters Date: Fri, 26 Jun 2026 10:23:26 +1000 Subject: [PATCH 09/10] Format polling timeout error message as a code block Co-Authored-By: Claude Opus 4.8 (1M context) --- .../tentacle/troubleshooting/troubleshooting-polling.mdx | 4 +++- 1 file changed, 3 insertions(+), 1 deletion(-) diff --git a/src/pages/docs/infrastructure/deployment-targets/tentacle/troubleshooting/troubleshooting-polling.mdx b/src/pages/docs/infrastructure/deployment-targets/tentacle/troubleshooting/troubleshooting-polling.mdx index f57414d77e..9311dbfa7b 100644 --- a/src/pages/docs/infrastructure/deployment-targets/tentacle/troubleshooting/troubleshooting-polling.mdx +++ b/src/pages/docs/infrastructure/deployment-targets/tentacle/troubleshooting/troubleshooting-polling.mdx @@ -112,7 +112,9 @@ If a Polling Tentacle is online but you occasionally see timeout errors during c **The Error:** -> A request was sent to a polling endpoint, but the polling endpoint did not collect the request within the allowed time (00:02:00), so the request timed out. +``` +A request was sent to a polling endpoint, but the polling endpoint did not collect the request within the allowed time (00:02:00), so the request timed out. +``` **The Cause:** From fb120c192ec6459bc450087b57637a020cd0c264 Mon Sep 17 00:00:00 2001 From: Luke Butters Date: Fri, 26 Jun 2026 10:33:38 +1000 Subject: [PATCH 10/10] Specify language on error message code block (MD040) Co-Authored-By: Claude Opus 4.8 (1M context) --- .../tentacle/troubleshooting/troubleshooting-polling.mdx | 2 +- 1 file changed, 1 insertion(+), 1 deletion(-) diff --git a/src/pages/docs/infrastructure/deployment-targets/tentacle/troubleshooting/troubleshooting-polling.mdx b/src/pages/docs/infrastructure/deployment-targets/tentacle/troubleshooting/troubleshooting-polling.mdx index 9311dbfa7b..2927f644c7 100644 --- a/src/pages/docs/infrastructure/deployment-targets/tentacle/troubleshooting/troubleshooting-polling.mdx +++ b/src/pages/docs/infrastructure/deployment-targets/tentacle/troubleshooting/troubleshooting-polling.mdx @@ -112,7 +112,7 @@ If a Polling Tentacle is online but you occasionally see timeout errors during c **The Error:** -``` +```text A request was sent to a polling endpoint, but the polling endpoint did not collect the request within the allowed time (00:02:00), so the request timed out. ```