BDR-4933 - cli/docs changes for v5.5.0 #5512
Conversation
Includes read scalability feature changes and others
smaher-edb
requested review from
Jagdish-Kewat-EDB
and removed request for
Jagdish-Kewat-EDB
| @@ -31,13 +40,17 @@ pgd create-proxy [flags] | |||
| ```text | |||
| --group-name string group name | |||
| -h, --help help for create-proxy | |||
| --proxy-mode string proxy mode (default, read-only, any); proxy will route connections to - | |||
There was a problem hiding this comment.
This proxy-mode is going to a thing of confusion as is...
the "default" should be write or write-leader or anything but "default" as that leads to having to answer questions like "What's the default?" "Default" "Yes, what is it?" and "Who's on first" :)
As I understand it, this is entirely an abstraction in pgd-cli (As the mode is actually determined by the states of the two listen port settings) so it should be low impact to fix it before shipping?
There was a problem hiding this comment.
Well, yes, that was what I had added in the initial version of RFC. The modes were - write, read-only and read-write (or any). But Petr specifically asked to not use word write anywhere because it creates confusion (because Write Leader is used for both Write and Read queries), and he mentioned he is kind of fed up of answering people who keeps asking same question whether Write Leader is only for Write (since v5.0.0).
We then discussed a lot on the naming (other option) but yes, no one had the better option. Current naming is kind of loosly based on target_session_attrs.
There was a problem hiding this comment.
As I understand it, this is entirely an abstraction in pgd-cli
Values are what SQL interface bdr.create_proxy() accepts. Hopefully, if this function goes away in future version (which is the plan) then this will also go away.
But yes, agree till that time it is going to be confusing. In fact, (I think I'd mentioned in our call also) not having proper mode field is going to be problem everywhere. Explaining in terms of port is odd.
There was a problem hiding this comment.
Therefore, on below line and in description have added the detail what each mode is.
There was a problem hiding this comment.
This option in both CLI cmd and sql interface is optional. So, hopefully, people will care about it only when they are trying to enable proxy for read-only connections.
There was a problem hiding this comment.
"follow" (the write leader) might have worked better.
There was a problem hiding this comment.
Hmm, not sure : ). "follow" in isolation may not look right.
@freza @ddipak do you have any opinion or suggestion here (as it will need change first in BDR). I don't mind changing it as we still have some time. But default is kind of also indicate it will work as it worked previously (which hopefully then we could tell that it means Write Leader) but yes 🤷.
There was a problem hiding this comment.
Oh, not suggesting changing it. Just we need to name things better (and giving an example)
| @@ -31,13 +40,17 @@ pgd create-proxy [flags] | |||
| ```text | |||
| --group-name string group name | |||
| -h, --help help for create-proxy | |||
| --proxy-mode string proxy mode (default, read-only, any); proxy will route connections to - | |||
There was a problem hiding this comment.
Well, yes, that was what I had added in the initial version of RFC. The modes were - write, read-only and read-write (or any). But Petr specifically asked to not use word write anywhere because it creates confusion (because Write Leader is used for both Write and Read queries), and he mentioned he is kind of fed up of answering people who keeps asking same question whether Write Leader is only for Write (since v5.0.0).
We then discussed a lot on the naming (other option) but yes, no one had the better option. Current naming is kind of loosly based on target_session_attrs.
| @@ -31,13 +40,17 @@ pgd create-proxy [flags] | |||
| ```text | |||
| --group-name string group name | |||
| -h, --help help for create-proxy | |||
| --proxy-mode string proxy mode (default, read-only, any); proxy will route connections to - | |||
There was a problem hiding this comment.
As I understand it, this is entirely an abstraction in pgd-cli
Values are what SQL interface bdr.create_proxy() accepts. Hopefully, if this function goes away in future version (which is the plan) then this will also go away.
But yes, agree till that time it is going to be confusing. In fact, (I think I'd mentioned in our call also) not having proper mode field is going to be problem everywhere. Explaining in terms of port is odd.
| @@ -31,13 +40,17 @@ pgd create-proxy [flags] | |||
| ```text | |||
| --group-name string group name | |||
| -h, --help help for create-proxy | |||
| --proxy-mode string proxy mode (default, read-only, any); proxy will route connections to - | |||
| default - Write Leader | |||
There was a problem hiding this comment.
here, and in description have tried to add the details.
| @@ -31,13 +40,17 @@ pgd create-proxy [flags] | |||
| ```text | |||
| --group-name string group name | |||
| -h, --help help for create-proxy | |||
| --proxy-mode string proxy mode (default, read-only, any); proxy will route connections to - | |||
There was a problem hiding this comment.
Therefore, on below line and in description have added the detail what each mode is.
| @@ -31,13 +40,17 @@ pgd create-proxy [flags] | |||
| ```text | |||
| --group-name string group name | |||
| -h, --help help for create-proxy | |||
| --proxy-mode string proxy mode (default, read-only, any); proxy will route connections to - | |||
There was a problem hiding this comment.
This option in both CLI cmd and sql interface is optional. So, hopefully, people will care about it only when they are trying to enable proxy for read-only connections.
|
Please don't edit here - this PR has been subsumed into DOCS-577 - the complete PGD 5.5 release branch. |
What Changed?
Includes read scalability feature changes and others