pawurb / ruby-pg-extras Goto Github PK

Hi, I'm not sure if this is the right place to file this (I'm actually using rails-pg-extras, however I think the exception I'm seeing is coming from this gem.

When loading the landing page for rails-pg-extras (diagnostics) it throws the following error:

An ActionView::Template::Error occurred in queries#index:

  undefined method `fetch' for nil:NilClass

        end.fetch("installed_version", false)
           ^^^^^^

I can confirm that the server does not have sslinfo installed and thus my theory is that this is tied to this:

If I navigate directly to the following url /pg_extras/queries?query_name=ssl_used I receive this error message:

PG::UndefinedFunction: ERROR: function ssl_is_used() does not exist LINE 3: SELECT ssl_is_used(); ^ HINT: No function matches the given name and argument types. You might need to add explicit type casts.

I'm having a hard time replicating it locally. Not sure if it makes sense that sslinfo really isn't installed?

Thanks for this awesome gem - I made a quick monkey patch to puts to return strings so that the output can be caputered/displayed differently. I think this is valid usecase to implement properly in the gem.

RubyPGExtras.puts_returns_string = true
# now all table prints return uncolorized strings

module RubyPGExtras
  @@puts_returns_string = false

  def self.puts_returns_string
    @@puts_returns_string
  end

  def self.puts_returns_string=(value)
    @@puts_returns_string = value
  end
end

class String
  REGEXP_PATTERN = /\033\[([0-9]+);([0-9]+);([0-9]+)m(.+?)\033\[0m|([^\033]+)/m

  def uncolorize
    self.scan(REGEXP_PATTERN).inject("") do |str, match|
      str << (match[3] || match[4])
    end
  end
end

module RubyPGExtras
  def self.puts(args)
    if @@puts_returns_string
      args.to_s.uncolorize
    else
      Kernel.puts args
    end
  end
end

module RubyPGExtras
  class DiagnosePrint
    def puts(args)
      RubyPGExtras.puts args
    end
  end
end

module RubyPGExtras
  class TableInfoPrint
    def puts(args)
      RubyPGExtras.puts args
    end
  end
end

module RubyPGExtras
  class IndexInfoPrint
    def puts(args)
      RubyPGExtras.puts args
    end
  end
end

Hi there!

I'm not a lawyer but colorize is under GPL 2.0 and it reminds me this : https://dev.to/cseeman/what-s-up-with-mimemagic-breaking-everything-he1

As alternative to colorize there is paint which is under MIT license and rainbow.

There's also an issue on colorize repo : fazibear/colorize#47

pg_buffercache is one of those built-in extensions that provides lots of good feedback. It provides a means for examining what's happening in the shared buffer cache in real time. I'm working on a few Rails tasks and I thought it would be a great idea to provide them here instead. Would you merge it if I come up with a pull request?

This is kind of evolving an idea used in null_indexes. Additionally to suggesting partial indexes when we have a lot of NULLs in the column, we can suggest partial indexes for columns where we have a high % of the same value.

For example, for the table with millions of users an index on role was added (to quickly find admins, for example). But 99% of the users are regular users, several are admins, moderators, etc. So the partial index where role != 'regular' would be beneficial.

This is a really helpful gem! I'm the author of GoodJob, which uses a lot of Advisory Locks. This gem (and rails-pg-extras) came up in an issue (bensheldon/good_job#480).

For Advisory Locks in all_locks, the age isn't accurate because the query is fetching the time of the connection's last query, not necessarily the age of the Advisory Lock itself:

I wanted to suggest renaming it, maybe last_activity_age (naming is hard).

And, I also wanted to suggest including pg_stat_activity.application_name in the output. I think that application_name can provide high-context information that would be really useful for understanding what's happening with the locks.

The gem has a dependency on the now unsupported filesize gem - https://github.com/dominikh/filesize

The dependency is extremely minor. It consists of two references

1. Conversion of 1 MB to a constant int value, which can just be replaced by the constant int value
2. Checking whether an index size is above a minimum, and hence should be analyzed in the unused_indexes case

Given that the dependency is now unsupported, it should probably be removed. I'm happy to put together a PR if there is interest in resolving the issue.

Didn't really have a better title for this but...

I've started building out a feature in our application that uses multiple schemas. This means, that when I later use rails-pg-extras and RailsPGExtras.index_usage I get indexes for materialized views I've created in this other schema. Since the views I'm creating have the same name as tables in the public schema, I'll get duplicate entries for users, for example. The output doesn't include the schema name so I don't get that information and can't pick the correct version of the index I care about.

Basically, the end goal is I want to include only the public schema for index_usage, and basically anything else that would include things in multiple schemas. So either I need to be able to customize the query, or get more information from the query so I can ignore it later on, or something else. I'm open to suggestions. I could fork the gem, but this seems like something that could be useful for other people.

Ideas? I can make a PR if there's some direction you think would be best.