Advice about index stuck

Question

I have not seen this case in the other topics: magento 1.9.2CE that has a reindex working for way too many hours seems stuck ; generally it takes few minutes to 1 hour.

---

Context

everyday at 02h10 a homemade script cronUpdateMagentoDaily.sh does stuff and then runs:

php5.6 /var/www/html/magento/shell/indexer.php --reindex catalog_product_price

---

Symptoms

• folder var/locks is empty
• process of reindexing still ongoing but time stuck at 0
• database shows an ended_at value strange

process_id indexer_code              status          started_at          ended_at            mode
         1 catalog_product_attribute pending         2019-09-12 03:11:01 2019-09-12 03:11:01 real_time
         2 catalog_product_price     working         2019-09-12 03:11:22 2019-09-12 03:11:01 real_time
         3 catalog_url               require_reindex 2019-09-02 15:52:08 2019-09-02 16:08:30 manual

Why would catalog_product_price have an ended_at value anterior to its started_at value? How can it be precisely the same as catalog_product_attribute ended_at value?

May that explain what that reindexation is stuck doing (apparently) nothing?

Running processes (snippet of ps -afux) showing that it does nothing:

USER       PID %CPU %MEM    VSZ   RSS TTY      STAT START   TIME COMMAND
root      3538  0.0  0.0  27732  1984 ?        Ss    2016   8:17 /usr/sbin/cron -f
root     30940  0.0  0.0  49044  2784 ?        S    02:10   0:00  \_ /usr/sbin/CRON -f
root     30954  0.0  0.0   4504   696 ?        Ss   02:10   0:00      \_ /bin/sh -c /var/www/html/magento/cronUpdateMagentoDaily.sh
root     30961  0.0  0.0  11244  2960 ?        S    02:10   0:00          \_ /bin/bash /var/www/html/magento/cronUpdateMagentoDaily.sh
root      7058  0.0  0.1 332556 46156 ?        S    05:11   0:00              \_ php5.6 /var/www/html/magento/shell/indexer.php --reindex catalog_product_price

May it be safe to kill its process and re-run it?

Answer 1

Since the time spent on CPU was still 0, I did the follonwing (in case it helps anyone):

1. kill the process (in my case, PID 7058)
   catalog_product_price is still at status "working"
2. look for anything in var/locks (in my case, still empty)
3. enforce the status in database: in table index_process set to "pending"
4. backend > Index managment : clic on the messed up index
5. set to manual update, save process
   catalog_product_price seems not to be locked anymore
6. run a reindex

Problem solved! Yet I have no idea about the origin.