API to collect list of running SQL queries

[Alexey Kuznetsov]

Hi!

Do we have API to collect list of running SQL queries?
If not, how hard is to implement this?


--
Alexey Kuznetsov

[Sergi]

List of SQL queries running where? On some exact server node? The original
SQL query is not available there, we only have generated MAP part of the
query.

Sergi

2016-12-15 6:13 GMT+03:00 Alexey Kuznetsov <[hidden email]>:

> Hi!
>
> Do we have API to collect list of running SQL queries?
> If not, how hard is to implement this?
>
>
> --
> Alexey Kuznetsov
>

[Vladimir Ozerov]

Sergi,

How do you think, will it affect performance significantly if we start
passing original SQL between nodes? Not only this will help Alexey, but
also improve user experience, because currently long-running queries are
printed in transformed form, and it is pretty hard for users to understand
what is going on.

Vladimir.

On Thu, Dec 15, 2016 at 11:28 AM, Sergi Vladykin <[hidden email]>
wrote:

> List of SQL queries running where? On some exact server node? The original
> SQL query is not available there, we only have generated MAP part of the
> query.
>
> Sergi
>
> 2016-12-15 6:13 GMT+03:00 Alexey Kuznetsov <[hidden email]>:
>
> > Hi!
> >
> > Do we have API to collect list of running SQL queries?
> > If not, how hard is to implement this?
> >
> >
> > --
> > Alexey Kuznetsov
> >
>

[Sergi]

I think in some cases it will not affect performance much (when query
processing time is big enough), but if it is a large enough query (large
string) which actually runs fast, this may affect throughput. I guess we
have to measure some scenarios and evaluate the possible performance drop
before making decisions.

Sergi

2016-12-15 11:57 GMT+03:00 Vladimir Ozerov <[hidden email]>:

> Sergi,
>
> How do you think, will it affect performance significantly if we start
> passing original SQL between nodes? Not only this will help Alexey, but
> also improve user experience, because currently long-running queries are
> printed in transformed form, and it is pretty hard for users to understand
> what is going on.
>
> Vladimir.
>
> On Thu, Dec 15, 2016 at 11:28 AM, Sergi Vladykin <[hidden email]
> >
> wrote:
>
> > List of SQL queries running where? On some exact server node? The
> original
> > SQL query is not available there, we only have generated MAP part of the
> > query.
> >
> > Sergi
> >
> > 2016-12-15 6:13 GMT+03:00 Alexey Kuznetsov <[hidden email]>:
> >
> > > Hi!
> > >
> > > Do we have API to collect list of running SQL queries?
> > > If not, how hard is to implement this?
> > >
> > >
> > > --
> > > Alexey Kuznetsov
> > >
> >
>

[Alexey Kuznetsov]

I created issue for this: https://issues.apache.org/jira/browse/IGNITE-4436

On Thu, Dec 15, 2016 at 4:11 PM, Sergi Vladykin <[hidden email]>
wrote:

> I think in some cases it will not affect performance much (when query
> processing time is big enough), but if it is a large enough query (large
> string) which actually runs fast, this may affect throughput. I guess we
> have to measure some scenarios and evaluate the possible performance drop
> before making decisions.
>
> Sergi
>
> 2016-12-15 11:57 GMT+03:00 Vladimir Ozerov <[hidden email]>:
>
>> Sergi,
>>
>> How do you think, will it affect performance significantly if we start
>> passing original SQL between nodes? Not only this will help Alexey, but
>> also improve user experience, because currently long-running queries are
>> printed in transformed form, and it is pretty hard for users to understand
>> what is going on.
>>
>> Vladimir.
>>
>> On Thu, Dec 15, 2016 at 11:28 AM, Sergi Vladykin <
>> [hidden email]>
>> wrote:
>>
>> > List of SQL queries running where? On some exact server node? The
>> original
>> > SQL query is not available there, we only have generated MAP part of the
>> > query.
>> >
>> > Sergi
>> >
>> > 2016-12-15 6:13 GMT+03:00 Alexey Kuznetsov <[hidden email]>:
>> >
>> > > Hi!
>> > >
>> > > Do we have API to collect list of running SQL queries?
>> > > If not, how hard is to implement this?
>> > >
>> > >
>> > > --
>> > > Alexey Kuznetsov
>> > >
>> >
>>
>
>

--
Alexey Kuznetsov

[Alexey Kuznetsov]

I like Vladimir idea.

May be we could *compress* large strings?
Or may be we could have API to ask for *string with original SQL* on any
node?

This *string with original SQL* only needed for special cases like
*monitoring utility* and *warning on long query*.


On Thu, Dec 15, 2016 at 4:11 PM, Sergi Vladykin <[hidden email]>
wrote:

> I think in some cases it will not affect performance much (when query
> processing time is big enough), but if it is a large enough query (large
> string) which actually runs fast, this may affect throughput. I guess we
> have to measure some scenarios and evaluate the possible performance drop
> before making decisions.
>
> Sergi
>
> 2016-12-15 11:57 GMT+03:00 Vladimir Ozerov <[hidden email]>:
>
>> Sergi,
>>
>> How do you think, will it affect performance significantly if we start
>> passing original SQL between nodes? Not only this will help Alexey, but
>> also improve user experience, because currently long-running queries are
>> printed in transformed form, and it is pretty hard for users to understand
>> what is going on.
>>
>> Vladimir.
>>
>> On Thu, Dec 15, 2016 at 11:28 AM, Sergi Vladykin <
>> [hidden email]>
>> wrote:
>>
>> > List of SQL queries running where? On some exact server node? The
>> original
>> > SQL query is not available there, we only have generated MAP part of the
>> > query.
>> >
>> > Sergi
>> >
>> > 2016-12-15 6:13 GMT+03:00 Alexey Kuznetsov <[hidden email]>:
>> >
>> > > Hi!
>> > >
>> > > Do we have API to collect list of running SQL queries?
>> > > If not, how hard is to implement this?
>> > >
>> > >
>> > > --
>> > > Alexey Kuznetsov
>> > >
>> >
>>
>
>

--
Alexey Kuznetsov

[dsetrakyan]

Alexey,

How about you only look at the client side, which obviously has the full
query?

D.

On Thu, Dec 15, 2016 at 1:35 AM, Alexey Kuznetsov <[hidden email]>
wrote:

> I like Vladimir idea.
>
> May be we could *compress* large strings?
> Or may be we could have API to ask for *string with original SQL* on any
> node?
>
> This *string with original SQL* only needed for special cases like
> *monitoring utility* and *warning on long query*.
>
>
> On Thu, Dec 15, 2016 at 4:11 PM, Sergi Vladykin <[hidden email]>
> wrote:
>
> > I think in some cases it will not affect performance much (when query
> > processing time is big enough), but if it is a large enough query (large
> > string) which actually runs fast, this may affect throughput. I guess we
> > have to measure some scenarios and evaluate the possible performance drop
> > before making decisions.
> >
> > Sergi
> >
> > 2016-12-15 11:57 GMT+03:00 Vladimir Ozerov <[hidden email]>:
> >
> >> Sergi,
> >>
> >> How do you think, will it affect performance significantly if we start
> >> passing original SQL between nodes? Not only this will help Alexey, but
> >> also improve user experience, because currently long-running queries are
> >> printed in transformed form, and it is pretty hard for users to
> understand
> >> what is going on.
> >>
> >> Vladimir.
> >>
> >> On Thu, Dec 15, 2016 at 11:28 AM, Sergi Vladykin <
> >> [hidden email]>
> >> wrote:
> >>
> >> > List of SQL queries running where? On some exact server node? The
> >> original
> >> > SQL query is not available there, we only have generated MAP part of
> the
> >> > query.
> >> >
> >> > Sergi
> >> >
> >> > 2016-12-15 6:13 GMT+03:00 Alexey Kuznetsov <[hidden email]>:
> >> >
> >> > > Hi!
> >> > >
> >> > > Do we have API to collect list of running SQL queries?
> >> > > If not, how hard is to implement this?
> >> > >
> >> > >
> >> > > --
> >> > > Alexey Kuznetsov
> >> > >
> >> >
> >>
> >
> >
>
>
> --
> Alexey Kuznetsov
>

[Alexey Kuznetsov]

Dmitriy,

Yes, that is what I'm going to do as first prototype.
But, I would prefer to discuss API with experts.

As Vladimir mentioned, there a usability use case with queries that also
could be discussed.

Sergi, could you, please, point a proper class where such API (for
collecting list of running queries) could be implemented?



On Thu, Dec 15, 2016 at 8:54 PM, Dmitriy Setrakyan <[hidden email]>
wrote:

> Alexey,
>
> How about you only look at the client side, which obviously has the full
> query?
>
> D.
>
> On Thu, Dec 15, 2016 at 1:35 AM, Alexey Kuznetsov <[hidden email]>
> wrote:
>
> > I like Vladimir idea.
> >
> > May be we could *compress* large strings?
> > Or may be we could have API to ask for *string with original SQL* on any
> > node?
> >
> > This *string with original SQL* only needed for special cases like
> > *monitoring utility* and *warning on long query*.
> >
> >
> > On Thu, Dec 15, 2016 at 4:11 PM, Sergi Vladykin <
> [hidden email]>
> > wrote:
> >
> > > I think in some cases it will not affect performance much (when query
> > > processing time is big enough), but if it is a large enough query
> (large
> > > string) which actually runs fast, this may affect throughput. I guess
> we
> > > have to measure some scenarios and evaluate the possible performance
> drop
> > > before making decisions.
> > >
> > > Sergi
> > >
> > > 2016-12-15 11:57 GMT+03:00 Vladimir Ozerov <[hidden email]>:
> > >
> > >> Sergi,
> > >>
> > >> How do you think, will it affect performance significantly if we start
> > >> passing original SQL between nodes? Not only this will help Alexey,
> but
> > >> also improve user experience, because currently long-running queries
> are
> > >> printed in transformed form, and it is pretty hard for users to
> > understand
> > >> what is going on.
> > >>
> > >> Vladimir.
> > >>
> > >> On Thu, Dec 15, 2016 at 11:28 AM, Sergi Vladykin <
> > >> [hidden email]>
> > >> wrote:
> > >>
> > >> > List of SQL queries running where? On some exact server node? The
> > >> original
> > >> > SQL query is not available there, we only have generated MAP part of
> > the
> > >> > query.
> > >> >
> > >> > Sergi
> > >> >
> > >> > 2016-12-15 6:13 GMT+03:00 Alexey Kuznetsov <[hidden email]>:
> > >> >
> > >> > > Hi!
> > >> > >
> > >> > > Do we have API to collect list of running SQL queries?
> > >> > > If not, how hard is to implement this?
> > >> > >
> > >> > >
> > >> > > --
> > >> > > Alexey Kuznetsov
> > >> > >
> > >> >
> > >>
> > >
> > >
> >
> >
> > --
> > Alexey Kuznetsov
> >
>

--
Alexey Kuznetsov