RDS custom oracle disk full

Question

Hello
We are currently using AWS RDS Custom Oracle
and the OS disk have a root partition of 10Gb , although allocating a 42 Gb volume . Just to add a cherry atop, there is a 16Gb swap area allocated in between, which make worse to expand to the remaining disk area

I could barely get out of os disk full condition killing old log files in /var/log

[root@ip- /]# df -kh
Filesystem                       Size  Used Avail Use% Mounted on
devtmpfs                         7.6G     0  7.6G   0% /dev
tmpfs                             16G  7.5G  7.8G  49% /dev/shm
tmpfs                            7.7G  785M  6.9G  11% /run
tmpfs                            7.7G     0  7.7G   0% /sys/fs/cgroup
/dev/nvme0n1p1                   9.8G  7.9G  1.8G  83% /
/dev/nvme1n1                      25G   13G   11G  54% /rdsdbbin
/dev/mapper/dbdata01-lvdbdata01  296G   25G  271G   9% /rdsdbdata
tmpfs                            1.6G     0  1.6G   0% /run/user/61001
tmpfs                            1.6G     0  1.6G   0% /run/user/61005

[root@ip- aws]# lsblk
NAME                    MAJ:MIN RM  SIZE RO TYPE MOUNTPOINT
nvme0n1                 259:0    0   42G  0 disk 
├─nvme0n1p1             259:1    0   10G  0 part /
├─nvme0n1p128           259:3    0    1M  0 part 
└─nvme0n1p127           259:2    0   16G  0 part [SWAP]
nvme3n1                 259:6    0   75G  0 disk 
└─dbn0-lvdbn0           252:0    0  300G  0 lvm  
  └─dbdata01-lvdbdata01 252:1    0  300G  0 lvm  /rdsdbdata
nvme2n1                 259:5    0   75G  0 disk 
└─dbn0-lvdbn0           252:0    0  300G  0 lvm  
  └─dbdata01-lvdbdata01 252:1    0  300G  0 lvm  /rdsdbdata
nvme5n1                 259:8    0   75G  0 disk 
└─dbn0-lvdbn0           252:0    0  300G  0 lvm  
  └─dbdata01-lvdbdata01 252:1    0  300G  0 lvm  /rdsdbdata
nvme1n1                 259:4    0   25G  0 disk /rdsdbbin
nvme4n1                 259:7    0   75G  0 disk 
└─dbn0-lvdbn0           252:0    0  300G  0 lvm  
  └─dbdata01-lvdbdata01 252:1    0  300G  0 lvm  /rdsdbdata

du -sh ./*/

185M	./bin/
143M	./boot/
7.5G	./dev/
43M	./etc/
124K	./home/
1.1G	./lib/
195M	./lib64/
16K	./lost+found/
4.0K	./media/
4.0K	./mnt/
3.4G	./opt/
0	./proc/
13G	./rdsdbbin/
25G	./rdsdbdata/
13M	./root/
785M	./run/
46M	./sbin/
4.0K	./srv/
0	./sys/
72K	./tmp/
465M	./usr/
2.5G	./var/

What im planning to do is to allocate a new swap volume , switch it on, kill the old swap and expand the original volume as the most i could.

1) Could this harm any monitoring task in AWS for RDS custom?

2)Any good soul at AWS could look into the automation scripts for RDS custom and make SWAP on a separate volume? , and allocate the 42Gb volume fully for OS?
There can happen some bad stuff any OS updates that surely need more disk space

Answer

Modifying the root volume or any volume for that matter will put you out of "support parameter" and automation will be paused. Your best bet is  be to open a service ticket and have product team involved from AWS side and they can provide suggestions as how to proceed.

Answer

Just a quick update

Thanks Dev

Ive did this totally online in the EC2 ...  Ive killed the swap apartition , and rezised the root partiton to the max volume size using parted (it reported some mismatched GPT table size, but fix option went well), and recreated the swap as file into the root partition
now the root partition have 42 Gb of space, which 25 is currently in use
ive had set up an external swap partition on another volume and yeah, the dreaded message unsupported flashed up. it went away a good chunk of time later 2-3hours after ive moved the swap to a file into the root partiton

CloudWatch and Log outputs were really cumbersome to use to troubleshoot why the message didnt went away when using a swap file. I could ask for a procedure or some script that you can run to manually detect whats off from the supported configuration. too much verbose to find the needle

RDS custom oracle disk full

相關內容