YDBOPS-11282 Some changes to make install more comfortable (#68)

* More stable port listener
Check for YDB presence on early stage

* Fix problem with tcp_htcp
+FAQ
Update list of supported OS

* Fix description

* Fix list of OS

---------

Co-authored-by: Eugene Arbatsky <[email protected]>

Author: elabpro

README.md

@@ -23,6 +23,17 @@ Current limitations:
 * the cluster configuration file has to be manually created;
 * there are no examples for configuring the storage nodes with different disk layouts (it seems to be doable by defining different `ydb_disks` values for different host groups).
 
+## Supported operation system
+
+- Ubuntu 20.04, 22.04, 24.04
+- Debian 11.11, 12.7
+- AstraLinux 1.7
+- AlmaLinux 8.9, 9.4, 9.5
+- Altlinux 8.4, 10, 10.1
+- RedHat 9.3
+- RedOS 7.3, 8
+- CentOS 8
+
 ## Ansible Collection - ydb_platform.ydb
 
 Documentation for the collection.
@@ -169,17 +180,6 @@ graph TD;
    end
 ```
 
-## Supported operation system
-
-- ydb_platform.ydb.initial_setup - Install cluster
-    - Ubuntu 22.04, 24.04
-    - Debian 11.11, 12.7
-    - AstraLinux 1.7
-    - AlmaLinux 8,9
-    - Altlinux 10
-    - RedHat 9
-    - RedOS 7
-
 # Install in isolated mode
 Isolated mode - situation when hosts are isolated from Internet (intranet, secure environment). There two possible way to install:
 1. Use bastion / jump host
@@ -277,3 +277,20 @@ sudo docker run -it --rm \
         -v /home/ansible/ydb-ansible:/root/.ansible/collections/ansible_collections/ydb_platform/ydb \
         ydb-ansible ansible-console ydb
 ```
+
+# FAQ
+1) Q: How to install on Linux with kernel 5.15.0-1073-kvm, which does not contain the tcp_htcp module?
+   A1: define empty variable `ydb_congestion_module` in inventory
+   A2: define variable in command line:
+   `ansible-playbook ydb_platform.ydb.initial_setup --extra-vars "ydb_congestion_module="`
+
+2) Q: How to handle error: `aborting playbook execution. Stop running YDB instances`?
+   A1: Manually stop YDB instances on the hosts for new YDB installation
+   A2: Use other hosts without YDB
+   A3: Use ansible-console to stop YDB instances:
+   ```
+   ansible-console ydb
+   $ sudo systemctl stop ydbd-storage
+   $ sudo systemctl stop ydbd-database-a
+   $ sudo systemctl stop ydbd-database-b
+   ```

roles/check_connectivity/files/port_listner.py

@@ -1,61 +1,80 @@
 # Simple script for checking connectivity
 # It's starting TCP server to listen several ports
+#
+# Expected result: Listen on IPv4/IPv6 interface for incoming requests
+# Version: 1.2
+# Author: Eugene Arbatsky
+#
 # Usage:
-#   python3 port_listner.py <port1> [..<portN>]
+#   python3 port_listner.py <FQDN> <port1> [..<portN>]
 #
 import socket
 import threading
 import sys
+import time
 
 def listen_port4(port):
-    try:
-        # Set up IPv4 address
-        s = socket.socket(socket.AF_INET, socket.SOCK_STREAM)
-        s.bind(("0.0.0.0", port))
-        s.listen(5)
+    tries = 0
+    while tries < 10:
+        try:
+            # Set up IPv4 address
+            s = socket.socket(socket.AF_INET, socket.SOCK_STREAM)
+            s.bind(("0.0.0.0", port))
+            s.listen(5)
 
-        print(f"Listen port IPv4 {port}...")
+            print(f"Listen port IPv4 {port}...")
 
-        while True:
-            conn, addr = s.accept()
-            with conn:
-                print(f"New connection from {addr}")
-            conn.close()
-    except:
-        print(f"No IPv4")
+            while True:
+                conn, addr = s.accept()
+                with conn:
+                    print(f"New connection from {addr}")
+                conn.close()
+            break
+        except:
+            print(f"No IPv4 or port is busy.. try again")
+        tries = tries + 1
+        time.sleep(1)
 
 def listen_port6(port):
-    try:
-        # Set up IPv6 address
-        s = socket.socket(socket.AF_INET6, socket.SOCK_STREAM)
-        s.bind(("::", port))
-        s.listen(5)
+    tries = 0
+    while tries < 10:
+        try:
+            # Set up IPv6 address
+            s = socket.socket(socket.AF_INET6, socket.SOCK_STREAM)
+            s.bind(("::", port))
+            s.listen(5)
 
-        print(f"Listen port IPv6 {port}...")
+            print(f"Listen port IPv6 {port}...")
 
-        while True:
-            conn, addr = s.accept()
-            with conn:
-                print(f"New connection from {addr}")
-            conn.close()
-    except:
-        print(f"No IPv6")
+            while True:
+                conn, addr = s.accept()
+                with conn:
+                    print(f"New connection from {addr}")
+                conn.close()
+            break
+        except:
+            print(f"No IPv6 or port is busy .. try again")
+        tries = tries + 1
+        time.sleep(1)
 
 
 def main():
-    if len(sys.argv) < 2:
+    if len(sys.argv) < 3:
         print("No ports to listen!")
         return
 
-    ports = [int(port) for port in sys.argv[1:]]
+    fqdn = sys.argv[1]
+    ports = [int(port) for port in sys.argv[2:]]
 
     threads = []
+    # Determine prefered IP protocol
+    info = socket.getaddrinfo(fqdn,ports[0])
 
     for port in ports:
-        thread = threading.Thread(target=listen_port4, args=(port,))
-        threads.append(thread)
-        thread.start()
-        thread = threading.Thread(target=listen_port6, args=(port,))
+        if 'AddressFamily.AF_INET6' in str(info[0][0]):
+            thread = threading.Thread(target=listen_port6, args=(port,))
+        else:
+            thread = threading.Thread(target=listen_port4, args=(port,))
         threads.append(thread)
         thread.start()
 

roles/check_connectivity/tasks/main.yaml

@@ -8,8 +8,27 @@
     group: root
     mode: 0755
 
+- name: Get PIDs of YDBD
+  shell: pgrep ydbd | wc -l
+  register: pids_of_ydbd
+
+- name: Check presence of YDBD
+  when: pids_of_ydbd.stdout != "0"
+  ansible.builtin.fail:
+    msg: "aborting playbook execution. Stop running YDB instances"
+
+- name: Stop script
+  shell: "pkill -f -e \"port_listner\""
+  ignore_errors: True
+
+- name: Additional delay for OS to release ports 
+  ansible.builtin.pause: seconds=5
+
 - name: Start script to listen YDB ports
-  shell: "nohup python3 /tmp/port_listner.py {{ ydb_ports| join(' ') }} &"
+  shell: "nohup python3 /tmp/port_listner.py {{ inventory_hostname }} {{ ydb_ports| join(' ') }} &"
+
+- name: Additional delay for OS to take ports
+  ansible.builtin.pause: seconds=5
 
 - name: Check connections between nodes
   wait_for:

roles/system/tasks/main.yaml

@@ -33,7 +33,7 @@
         mode: '0644'
     - name: Load the {{ ydb_congestion_module }} module
       command: "/sbin/modprobe {{ ydb_congestion_module }}"
-  when: ydb_congestion_module is defined and ansible_distribution != "RedHat"
+  when: ydb_congestion_module is defined and ydb_congestion_module != '' and ansible_distribution != "RedHat"
 
 - name: Refresh the current sysctl settings
   command: "/sbin/sysctl --system"