Bokeh (+ Dependencies) is quite large

Benji_Xie · January 28, 2015, 6:17pm

Not really a problem or a question but more of an FYI:

**Bokeh (0.7.0) is quite large. **

I have been trying to host a dashboard built with bokeh widgets (similar to stock_applet). Thus far, I have tried both Heroku and OpenShift.

The issue I have been running into is that the services I have been using time out when attempting to include Bokeh and its dependencies (usually around building numpy).

With Heroku, it is a 15 minute timeout (you can get around this by submitting a ticket and asking for additional build time).

With Openstack, the error message is a bit more cryptic, but I also suspect it is a timeout.

Again, just an FYI. If anyone has found a (free) PaaS that will accept Bokeh, please let me know!

Nick_Roth · January 28, 2015, 6:22pm

When you say “Bokeh”, do you mean the python side, javascript side, or both?

Also, building numpy? For those services, why not just use the pre-built stuff from conda? I haven’t worked with Heroku and conda, but a quick search popped up with this: https://github.com/kennethreitz/conda-buildpack/blob/master/README.md

···

On Wednesday, January 28, 2015 at 12:17:24 PM UTC-6, Benji Xie wrote:

Not really a problem or a question but more of an FYI:

**Bokeh (0.7.0) is quite large. **

I have been trying to host a dashboard built with bokeh widgets (similar to stock_applet). Thus far, I have tried both Heroku and OpenShift.

The issue I have been running into is that the services I have been using time out when attempting to include Bokeh and its dependencies (usually around building numpy).

With Heroku, it is a 15 minute timeout (you can get around this by submitting a ticket and asking for additional build time).

With Openstack, the error message is a bit more cryptic, but I also suspect it is a timeout.

Again, just an FYI. If anyone has found a (free) PaaS that will accept Bokeh, please let me know!

Bryan · January 28, 2015, 6:26pm

Hi Benji,

I am not really familiar with Heroku, it sounds like you are trying to build everything form source? If possible I would highly recommend deploying with conda if you can. You can do:

conda install bokeh

and that will grab (precompiled, binary) packages for bokeh and all its dependencies that should install in seconds.

Bryan

···

On Jan 28, 2015, at 12:17 PM, Benji Xie <[email protected]> wrote:

Not really a problem or a question but more of an FYI:

Bokeh (0.7.0) is quite large.

I have been trying to host a dashboard built with bokeh widgets (similar to stock_applet). Thus far, I have tried both Heroku and OpenShift.
The issue I have been running into is that the services I have been using time out when attempting to include Bokeh and its dependencies (usually around building numpy).

With Heroku, it is a 15 minute timeout (you can get around this by submitting a ticket and asking for additional build time).
With Openstack, the error message is a bit more cryptic, but I also suspect it is a timeout.

Again, just an FYI. If anyone has found a (free) PaaS that will accept Bokeh, please let me know!

--
You received this message because you are subscribed to the Google Groups "Bokeh Discussion - Public" group.
To unsubscribe from this group and stop receiving emails from it, send an email to [email protected].
To post to this group, send email to [email protected].
To view this discussion on the web visit https://groups.google.com/a/continuum.io/d/msgid/bokeh/86310298-90c7-4e19-9650-dba7bceda6e1%40continuum.io\.
For more options, visit https://groups.google.com/a/continuum.io/d/optout\.

Benji_Xie · January 28, 2015, 6:27pm

Python side. I’ve been trying to include bokeh as a dependency (in requirements.txt).
This buildpack is interesting/exciting…

···

On Wednesday, January 28, 2015 at 1:22:28 PM UTC-5, Nick Roth wrote:

When you say “Bokeh”, do you mean the python side, javascript side, or both?

Also, building numpy? For those services, why not just use the pre-built stuff from conda? I haven’t worked with Heroku and conda, but a quick search popped up with this: https://github.com/kennethreitz/conda-buildpack/blob/master/README.md

On Wednesday, January 28, 2015 at 12:17:24 PM UTC-6, Benji Xie wrote:

Not really a problem or a question but more of an FYI:

**Bokeh (0.7.0) is quite large. **

I have been trying to host a dashboard built with bokeh widgets (similar to stock_applet). Thus far, I have tried both Heroku and OpenShift.

The issue I have been running into is that the services I have been using time out when attempting to include Bokeh and its dependencies (usually around building numpy).

With Heroku, it is a 15 minute timeout (you can get around this by submitting a ticket and asking for additional build time).

With Openstack, the error message is a bit more cryptic, but I also suspect it is a timeout.

Again, just an FYI. If anyone has found a (free) PaaS that will accept Bokeh, please let me know!

Sarah_Bird3 · January 28, 2015, 7:40pm

Hi all,

Some notes from my efforts to get bokeh working on heroku (which I have now).

Firstly, I used a buildpack:

$ heroku config:set BUILDPACK_URL=[https://github.com/thenovices/heroku-buildpack-scipy](https://github.com/thenovices/heroku-buildpack-scipy)

I needed to add numpy, pandas, bokeh slowly, so build wouldn’t time out. So:

I added numpy to my requirements.txt, and deployed.
Then added pandas to my requirements.txt, and deployed.
Then added bokeh to my requirements.txt, and deployed.

Each time it checks whether packages are already installed so this incremental approach worked. I can imagine this could get tedious, but worked for me.

Finally, at some point, can’t remember which, I needed to do:

$ heroku config:add LD_LIBRARY_PATH=/app/.heroku/vendor/lib/atlas-base/atlas:/app/.heroku/vendor/lib/atlas-base:/app/.heroku/vendor/lib/

(from: https://stackoverflow.com/questions/11635857/how-to-install-scikit-learn-on-heroku-cedar)

Hope this helps,

Bird

···

On Wed, Jan 28, 2015 at 10:27 AM, Benji Xie [email protected] wrote:

Python side. I’ve been trying to include bokeh as a dependency (in requirements.txt).
This buildpack is interesting/exciting…

On Wednesday, January 28, 2015 at 1:22:28 PM UTC-5, Nick Roth wrote:

When you say “Bokeh”, do you mean the python side, javascript side, or both?

Also, building numpy? For those services, why not just use the pre-built stuff from conda? I haven’t worked with Heroku and conda, but a quick search popped up with this: https://github.com/kennethreitz/conda-buildpack/blob/master/README.md

On Wednesday, January 28, 2015 at 12:17:24 PM UTC-6, Benji Xie wrote:

Not really a problem or a question but more of an FYI:

**Bokeh (0.7.0) is quite large. **

I have been trying to host a dashboard built with bokeh widgets (similar to stock_applet). Thus far, I have tried both Heroku and OpenShift.

The issue I have been running into is that the services I have been using time out when attempting to include Bokeh and its dependencies (usually around building numpy).

With Heroku, it is a 15 minute timeout (you can get around this by submitting a ticket and asking for additional build time).

With Openstack, the error message is a bit more cryptic, but I also suspect it is a timeout.

Again, just an FYI. If anyone has found a (free) PaaS that will accept Bokeh, please let me know!

–

You received this message because you are subscribed to the Google Groups “Bokeh Discussion - Public” group.

To unsubscribe from this group and stop receiving emails from it, send an email to [email protected].

To post to this group, send email to [email protected].

To view this discussion on the web visit https://groups.google.com/a/continuum.io/d/msgid/bokeh/4153fe35-85f4-491d-9465-0ab059188acb%40continuum.io.

For more options, visit https://groups.google.com/a/continuum.io/d/optout.

Benji_Xie · January 28, 2015, 8:22pm

Conda backpack works perfectly! Built numpy, pandas, and bokeh in one push!

The readme for the repo is quite nice, but more detailed instructions here (just remember to reference the original backpack in BACKPACK_URL)

···

On Wednesday, January 28, 2015 at 2:41:05 PM UTC-5, Sarah Bird wrote:

Hi all,

Some notes from my efforts to get bokeh working on heroku (which I have now).

Firstly, I used a buildpack:

$ heroku config:set BUILDPACK_URL=[https://github.com/thenovices/heroku-buildpack-scipy](https://github.com/thenovices/heroku-buildpack-scipy)

I needed to add numpy, pandas, bokeh slowly, so build wouldn’t time out. So:

I added numpy to my requirements.txt, and deployed.

Then added pandas to my requirements.txt, and deployed.

Then added bokeh to my requirements.txt, and deployed.

Each time it checks whether packages are already installed so this incremental approach worked. I can imagine this could get tedious, but worked for me.

Finally, at some point, can’t remember which, I needed to do:

$ heroku config:add LD_LIBRARY_PATH=/app/.heroku/vendor/lib/atlas-base/atlas:/app/.heroku/vendor/lib/atlas-base:/app/.heroku/vendor/lib/
(from: https://stackoverflow.com/questions/11635857/how-to-install-scikit-learn-on-heroku-cedar)

Hope this helps,

Bird

On Wed, Jan 28, 2015 at 10:27 AM, Benji Xie [email protected] wrote:

Python side. I’ve been trying to include bokeh as a dependency (in requirements.txt).
This buildpack is interesting/exciting…

On Wednesday, January 28, 2015 at 1:22:28 PM UTC-5, Nick Roth wrote:

When you say “Bokeh”, do you mean the python side, javascript side, or both?

Also, building numpy? For those services, why not just use the pre-built stuff from conda? I haven’t worked with Heroku and conda, but a quick search popped up with this: https://github.com/kennethreitz/conda-buildpack/blob/master/README.md

On Wednesday, January 28, 2015 at 12:17:24 PM UTC-6, Benji Xie wrote:

Not really a problem or a question but more of an FYI:

**Bokeh (0.7.0) is quite large. **

I have been trying to host a dashboard built with bokeh widgets (similar to stock_applet). Thus far, I have tried both Heroku and OpenShift.

The issue I have been running into is that the services I have been using time out when attempting to include Bokeh and its dependencies (usually around building numpy).

With Heroku, it is a 15 minute timeout (you can get around this by submitting a ticket and asking for additional build time).

With Openstack, the error message is a bit more cryptic, but I also suspect it is a timeout.

Again, just an FYI. If anyone has found a (free) PaaS that will accept Bokeh, please let me know!

–

You received this message because you are subscribed to the Google Groups “Bokeh Discussion - Public” group.

To unsubscribe from this group and stop receiving emails from it, send an email to [email protected].

To post to this group, send email to [email protected].

To view this discussion on the web visit https://groups.google.com/a/continuum.io/d/msgid/bokeh/4153fe35-85f4-491d-9465-0ab059188acb%40continuum.io.

For more options, visit https://groups.google.com/a/continuum.io/d/optout.

Sarah_Bird3 · January 28, 2015, 9:09pm

That’s really good to know. Thanks for reporting back.

···

On Wed, Jan 28, 2015 at 12:22 PM, Benji Xie [email protected] wrote:

Conda backpack works perfectly! Built numpy, pandas, and bokeh in one push!

The readme for the repo is quite nice, but more detailed instructions here (just remember to reference the original backpack in BACKPACK_URL)

On Wednesday, January 28, 2015 at 2:41:05 PM UTC-5, Sarah Bird wrote:

Hi all,

Some notes from my efforts to get bokeh working on heroku (which I have now).

Firstly, I used a buildpack:

$ heroku config:set BUILDPACK_URL=[https://github.com/thenovices/heroku-buildpack-scipy](https://github.com/thenovices/heroku-buildpack-scipy)

I needed to add numpy, pandas, bokeh slowly, so build wouldn’t time out. So:

I added numpy to my requirements.txt, and deployed.

Then added pandas to my requirements.txt, and deployed.

Then added bokeh to my requirements.txt, and deployed.

Each time it checks whether packages are already installed so this incremental approach worked. I can imagine this could get tedious, but worked for me.

Finally, at some point, can’t remember which, I needed to do:

$ heroku config:add LD_LIBRARY_PATH=/app/.heroku/vendor/lib/atlas-base/atlas:/app/.heroku/vendor/lib/atlas-base:/app/.heroku/vendor/lib/
(from: https://stackoverflow.com/questions/11635857/how-to-install-scikit-learn-on-heroku-cedar)

Hope this helps,

Bird

On Wed, Jan 28, 2015 at 10:27 AM, Benji Xie [email protected] wrote:

Python side. I’ve been trying to include bokeh as a dependency (in requirements.txt).
This buildpack is interesting/exciting…

On Wednesday, January 28, 2015 at 1:22:28 PM UTC-5, Nick Roth wrote:

When you say “Bokeh”, do you mean the python side, javascript side, or both?

Also, building numpy? For those services, why not just use the pre-built stuff from conda? I haven’t worked with Heroku and conda, but a quick search popped up with this: https://github.com/kennethreitz/conda-buildpack/blob/master/README.md

On Wednesday, January 28, 2015 at 12:17:24 PM UTC-6, Benji Xie wrote:

Not really a problem or a question but more of an FYI:

**Bokeh (0.7.0) is quite large. **

I have been trying to host a dashboard built with bokeh widgets (similar to stock_applet). Thus far, I have tried both Heroku and OpenShift.

The issue I have been running into is that the services I have been using time out when attempting to include Bokeh and its dependencies (usually around building numpy).

With Heroku, it is a 15 minute timeout (you can get around this by submitting a ticket and asking for additional build time).

With Openstack, the error message is a bit more cryptic, but I also suspect it is a timeout.

Again, just an FYI. If anyone has found a (free) PaaS that will accept Bokeh, please let me know!

–

You received this message because you are subscribed to the Google Groups “Bokeh Discussion - Public” group.

To unsubscribe from this group and stop receiving emails from it, send an email to [email protected].

To post to this group, send email to [email protected].

To view this discussion on the web visit https://groups.google.com/a/continuum.io/d/msgid/bokeh/4153fe35-85f4-491d-9465-0ab059188acb%40continuum.io.

For more options, visit https://groups.google.com/a/continuum.io/d/optout.

–

You received this message because you are subscribed to the Google Groups “Bokeh Discussion - Public” group.

To unsubscribe from this group and stop receiving emails from it, send an email to [email protected].

To post to this group, send email to [email protected].

To view this discussion on the web visit https://groups.google.com/a/continuum.io/d/msgid/bokeh/ea7947bc-be95-4ad8-b652-814752feb8cf%40continuum.io.

For more options, visit https://groups.google.com/a/continuum.io/d/optout.