How to copy a couchdb in dokku - couchdb

Question

I want to take a snapshot of one database running in app www and put it into app staging. When I do that with clone or create/import none of the data is available.
How am I meant to do it?
matt#server:~$ dokku run www curl http://www:password#dokku-couchdb-www:5555/www
{"db_name":"www","doc_count":4966,"doc_del_count":232,"update_seq":46475,"purge_seq":0,"compact_running":false,"disk_size":3071180923,"data_size":334987077,"instance_start_time":"1500006610823893","disk_format_version":6,"committed_update_seq":46475}
So from that you can see there are 4966 documents.
matt#server:~$ dokku couchdb:clone www staging_www
-----> Starting container
Waiting for container to be ready
=====> CouchDB container created: staging_www
DSN: http://staging_www:password#dokku-couchdb-staging-www:5555/staging_www
-----> Copying data from www to staging_www
% Total % Received % Xferd Average Speed Time Time Time Current
Dload Upload Total Spent Left Speed
100 1110M 0 1110M 0 0 30.4M 0 --:--:-- 0:00:36 --:--:-- 31.9M
-----> Done
So there are no errors in the clone. Then I run
dokku couchdb:link staging_www staging
dokku couchdb:promote staging_www staging
And there are no errors, but if I check the DB:
matt#server:~$ dokku run staging curl http://staging_www:password#dokku-couchdb-staging-www:5555/staging_www
{"db_name":"staging_www","doc_count":1,"doc_del_count":0,"update_seq":1,"purge_seq":0,"compact_running":false,"disk_size":4188,"data_size":342,"instance_start_time":"1509536857606369","disk_format_version":6,"committed_update_seq":1}
The doc count is 1 and I can't access any of the data in the staging app.
Equally I have tried
dokku couchdb:export www > www.couch
dokku couchdb:create staging_www
dokku couchdb:import staging_www < www.couch
dokku couchdb:link staging_www staging
dokku couchdb:promote staging_www staging
There are no errors, but again I end up with 1 doc in the database.
What am I meant to do?

Answer 1

With dokku 0.9.4 and 'dokku couchdb service plugin' 1.0.0
Solution is quite simple. After the first attempt
root#dokku01:~# dokku couchdb:clone www staging_www
and fail to clone db. You need to destroy staging_www
root#dokku01:~# dokku couchdb:destroy staging_www
and do cloning again.
root#dokku01:~# dokku couchdb:clone www staging_www
Now it will works as expected. You can check new db with
root#dokku01:~# curl -X GET 'http://staging_www:password#dokku-couchdb-staging-www:5555/staging_www/_all_docs?include_docs=true&attachments=true'
Exporting from www and then importing a dump in newly created staging_www also works.
It's a bug in CouchDB plugin and it will be quite fun to find what is the root cause of it.
UPDATE
The root cause of this bug is bash script for backup and restore 'couchdb-backup'. In some cases a 'curl calls' inside the script fail to reach the db and consequently a restore operation doesn't work.
Cloning db is: create new instance, export (backup) old data and then import (restore) that backup in the new db.
Script below is updated 'clone' script from couchdb service plugin. It's made more resilient to 'couchdb-backup' import quirks.
#!/usr/bin/env bash
source "$(dirname "$(cd "$(dirname "${BASH_SOURCE[0]}")" && pwd)")/config"
set -eo pipefail; [[ $DOKKU_TRACE ]] && set -x
source "$PLUGIN_BASE_PATH/common/functions"
source "$(dirname "$(cd "$(dirname "${BASH_SOURCE[0]}")" && pwd)")/functions"
service-clone-cmd() {
#E you can clone an existing service to a new one
#E dokku $PLUGIN_COMMAND_PREFIX:clone lolipop lolipop-2
#A service, service to run command against
#A new-service, name of new service
declare desc="create container <new-name> then copy data from <name> into <new-name>"
local cmd="$PLUGIN_COMMAND_PREFIX:clone" argv=("$#"); [[ ${argv[0]} == "$cmd" ]] && shift 1
declare SERVICE="$1" NEW_SERVICE="$2" CLONE_FLAGS_LIST="${#:3}"
[[ -z "$SERVICE" ]] && dokku_log_fail "Please specify a name for the service"
[[ -z "$NEW_SERVICE" ]] && dokku_log_fail "Please specify a name for the new service"
verify_service_name "$SERVICE"
PLUGIN_IMAGE=$(service_version "$SERVICE" | grep -o "^.*:" | sed -r "s/://g")
PLUGIN_IMAGE_VERSION=$(service_version "$SERVICE" | grep -o ":.*$" | sed -r "s/://g")
service_create "$NEW_SERVICE" "${#:3}"
dokku_log_info1 "Copying data from $SERVICE to $NEW_SERVICE"
attempts=5
attemptcount=0
R=2
succ_str=' Successfully.'
until [[ $R = 0 || $R = 1 ]]; do
attemptcount=$((attemptcount+1))
STDOUT1=$(service_export "$SERVICE" | service_import "$NEW_SERVICE" 2>&1) || true
if [[ ! "$STDOUT1" = *"${succ_str}" ]]; then
if [ $attemptcount = $attempts ]; then
R=1
echo -e "\nERROR: CouchDB Import failed - Stopping\n"
else
echo -e "\nWARN: CouchDB Import Reported an error - Attempt ${attemptcount}/${attempts} - Retrying...\n"
sleep 1
fi
else
R=0
fi
done
dokku_log_info1 "Done"
exit $R
}
service-clone-cmd "$#"
Old version at /var/lib/dokku/plugins/available/couchdb/subcommands/clone should be replaced with this new one.

Answer 2

I usually just use the couchdb dokku plugin found here
Curl can have a tendency to timeout with multiple files, so I use the plugin to stage. Easy peasy