Misc

Terminating bgp with apipa blocks on firewalls - Avoid at all times -

The problem

[Day 1]

Today i got lost in sophos firewalls internal routing and nat labyrinths, what i was trying to do was a simple LDAP integration to a server ath the end of a vpn tunnel. For interoperability reasons with AWS networks, we did used the famous "APIPA" block ip addresses on our bgp neigborship design.

But, there is catch !

There is always a catch and sometimes a group of catches. Windows server machines won't route or process apipa address ranges. Блять right ? Anyway my options were clear

  1. Change the addressing on ipsec links
  2. Snat the connections getting out of the firewall

Here is the topology i was dealing with.

i got close

I thought that amazon must have a sound reason to use these blocks on their integrations. I am sure they never thought the apipa addresses were going to be used as source addresses on production links, or the CPE's might have limitations too.

I know that Fortigates have the option to specify source addresses for control plane operations however sophos doesn't have them. _insert sad emoji here _

there is one

Ok, i can always use the snat option right ? So i went with option 2 on my list.

And after two hours mangling with it i gave up and created a case, which got the required attention immediately from sophos, since the traffic is generated from firewall itself, a special snat entry was not going to be processed as i expect.

Apparently Postrouting operations weren't processed on control plane chains, and i really don't wanted to go with option 2 so a very capable and gentle guy from sophos tried helped me however we couldn't got the golden ticket today !

I did spent 3 hours working on this case learned many things about how kernel processes encrypted packets and found out good documents about how packets traverse when they are using xfrm framework. i am going to add the things i read at #further reading section below.

And will update this page on next update i get.

Further Reading

Beautiful explanation of xfrm

How to use snat with xfrm if had a vyos

This image from the post above displays the packet flow mech on linux.

Strongswan xfrm implementation

The solution

[day 5]

Apparently, there is a special section to add nat entries for "System Generated Traffic", and it is only accessible to command line interface. To do that you have to access the firewall through console using ssh and go to advanced menu as shown below.

Then you will have to type

cish

then you'll have to add a nat entry as shown below

set advanced-firewall sys-traffic-nat add destination 172.17.21.11 netmask 255.255.255.255 snatip 172.17.24.1

[](https://books.netdev.com.tr/uploads/images/gallery/2022-12/image-1672316296662.png)

After you ping the destination or check the firewall you will see that the traffic is natted.

Wazuh Cheat Sheet

install linux agent on SLES or Opensuse

To install

WAZUH_MANAGER='172.21.21.165' WAZUH_AGENT_GROUP='Linux_Servers' zypper --no-gpg-checks install https://packages.wazuh.com/4.x/yum/wazuh-agent-4.3.10-1.x86_64.rpm

to check

systemctl status wazuh-agent.service

output

Gns3 as a daemon on 22.04

Kullanıcı ekliyoruz

sudo adduser gns3

Gerekli python libraryler

sudo apt install python3-setuptools python3-aiohttp python3-psutil python3-jsonschema git -y

Tutundurma süreci

Toplantı öncesi bitmesi gerekenler

  1. Sözleşmeler gönderilir (Hizmet ve Gizlilik Sözleşmesi)
  2. Tüm çalışanlarının mail adresleri alınıp ‘’Bilgilendirme’’ maili atılır (OKAN)
  3. Sistem sorumlusu ve Şirket sahibi/Genel müdür bilgileri öğrenilir.
  4. Envanter için personel gönderilir ve detaylı envanter çıkarılır. Envanter alınırken sistem analizi ön çalışması yapılır.
  5. Mutlaka Android telefon ile PHING kaydı alınır. Kritik önemdeki şifreler random pass yöntemi ile hemen değiştirilir.

Tanışma toplantısı

  1. Sistem sorumlusu ve veya Genel Müdür seviyesinde randevu talep edilir ve aşağıdaki konular görüşülür.

     1.1 Değerleme/analiz çalışması için ikinci ziyaret yapılır. (Müşteri ilişkileri yöneticisi)
 1.2 İş Kritik saat ve günler öğrenilir - En yoğun günleriniz hangileridir -
 1.3 İş Kritik sistemlerin duruşa - Asla durmaması gereken sistemleriniz nelerdir -
 1.4 Tek nokta problem analizi yapılır - sadece 1 adet sunucumu var ? firewall sadece 1 tane mi ?-
 1.5 Dışarıdan erişim politikaları öğrenilir -SSL vpn kullanıyormusunuz ?-
 1.6.En sık yaşanılan sıkıntılar konusunda bilgi talep edilir -En sık yaşadığınız problem nedir-
 1.7 Sık yaşanan konular için akla gelen çözümler var ise basitçe sıralanır, case olarak üretmek üzere toplantı sonrası adımlarında kullanılır
 1.8 Şirketimizin geçmişinden sayılarla örnekler verilir 

     a. Kısa tarihçe anlatılır - şu kadar yıldır varız, şu kadar kişiyiz - 
     b. Haftalık case ve çözülen iş sayısı söylenir
     c. 2.000 problem çözdük, 1800 işi tam zamamnında 150 işi yarı gecikmeli yaptık şeklinde istatistik verilir

  2. Müşteri den Yesis yetkilisi ataması yapılması istenir.

     6.1 Şirket bilgisayarından mutlaka Yesis login olduğu görülür
 6.2 Yesis konusunda kısaca bilgilendirilir.

Toplantı sonrası yapılması gerekenler

  1. Müşteriye, yesiscrm.net/mustericm paneli açılır

     6.3 İlk case oluşturması sağlanır. bu adında sık yaşanan problemler aşamasında alınan notlar kullanılır
 6.4 Önceden haber verilen ekibin case'e hızlı dönmesi sağlanır
 6.5 Bu konuda teşvik edici sözler söylenir -Ekran üzerinden yapılan işlemler hızlı dönüş sürecini başlatır-

  2. Teknik personellerin zabbix kayıt işlemleri kontrol edilir.

  3. Tespit edilen noksanlıklar raporlanır;

  4. Alınması gereken tedbirler ve çözümler oluşturulur.

  5. Gerekirse tedbir veya çözümlerin üretilmesi için teklifler oluşturulur.

  6. Sistem analizi ve diagram ile birlikte müşteri sunuma gidilip eksikleri ve son durumu iletilir.

  7. Sunum toplantısından sonra risk analiz raporunda tespit edilen acil durum çözümleri için teklif iletilmesi sağlanır(satış personeli)

  8. 2 ayda 1 Periyodik memnuniyet ziyareti takvimi oluşturulur (Müşteri ilişkileri Yöneticisi)

tutundurma.xmind

SDS

1. Bölüm

Storage Nedir

Diskler, cdler, usb stickler, floppy veri yazdığımız her media bir storage dır. bunları birşekilde birleştirip kocaman veri ambarları yapıyoruz, bunlara da depolama çözümü deniyor yani storage solution. Bunların donanım şeklinde çalışanı var yazılım şeklinde çalışanı var. Fiyat/performans hesabına göre trendyol gibi bir skalada donanımsal çözümün hem finansal hemde politik bazı kötü sonuçları oluyor bu sebeple yazılım tabanlı depolama çözümleri kullanmak daha doğru oluyor.

Software defined storage nedir

Yazılım tanımlı depolama, depolama yazılımını donanım platformundan ayırarak depolama kaynaklarını soyutlayan bir depolama mimarisidir. Bu sayede daha büyük esneklik, verimlilik ve ölçeklenebilirlik elde edilir. openstack in support ettikleri:

Bizim openstack teki hangisidir ?

Block storage için Ceph'in mimic versiyonu kullanıyoruz. Ana storage seçimimiz bu.

Biraz rakam verelim şu andaki büyüklüğü vs vs

Bunları neden anlatıyoruz ?

B ir sonraki slayta io konuştuktan sonra bağlayacağız

2. Bölüm

IO nedir

Yazma ve okuma isteklerinin tamamına denir, en önemli parametresi gecikme takip eden parametre ise süratidir ve iops değeri ile ifade edilir.

Bizim ceph üzerinden verdiğimiz io rate nedir ve neden böyledir

Excel üzerinden gösterelim

Bağlayalım

3. Bölüm

Sorun dan bahsettik, cevabımız "Ephemeral" nedir ?

Sözlük anlamı geçiçi geçici çünkü hostu yada diski kaybedersek buna cevabımız bazen biraz beklemek, bazen de yeni makina vermek olabiliyor.

Ne zaman kullanmalıyız ?

4. Production daki etkisi

5. Bölüm

Nopasswd the user

On ubuntu 22.04

to add your current user to passwordless sudo

echo "useradmin ALL=(ALL) NOPASSWD: ALL" > /etc/sudoers.d/useradmin

Freeradius 2fa with ldap and google

root@zabbix:/usr/local/etc/raddb/sites-enabled# more my_server

server my_server {
listen {
        type = auth
        ipaddr = *
        port = 1812
}
authorize {
        filter_uuid
        filter_google_otp
        ldap
        if (ok || updated)  {
        update control {
        Auth-Type := LDAP
        }
        }
}
authenticate {
        Auth-Type LDAP {
                ldap
        }
}
}

ldap conf

ldap {
   
    identity = 'CN=freeradius_svc,OU=Service_accounts,DC=migrosonline,DC=com'
    password = 'Cumartesi2023.'

  
    base_dn = 'OU=ADoutVPNusers,DC=migrosonline,DC=com'
    
    user {
         base_dn = 'OU=ADoutVPNusers,DC=migrosonline,DC=com'
         filter = "(sAMAccountName=%{%{Stripped-User-Name}:-%{User-Name}})" 
         control:My_Group = 'memberOf' 
         }


  
  }

Bgp Evpn vxlan Case Documentation

This is a lab to test your bgp/evpn/vxlan skills, You have 15 business days to finish this.

The case will be reviewed at a remote meeting session upon your declaration about golden tickets. No manual config is allowed, everything should be deployed using ansible-playbooks

You can access the lab at:

http://gns3.eclit.cloud:3080

You have to inform Eclit network team using e-mail opr-network@eclit.com before you start working, lab will be ready approximately 30 minuters after receving a confirmation e-mail from eclit.

There are other labs available, you will use the one with the name "your_name_network_case_eclit"

Facts

Scenario

You will need to deliver the requirements listed below

Using ansible playbooks:

Monitoring:

Golden tickets:

  1. vcs machines should be able to ping each other on same vlan between racks.
  2. A monitoring solution, it must update the operator about thresolds and outages. These updates must happen less than a minute when any sensor triggers.
  3. A solution brief, one pager.

100 kb over 20ms delayed link with smbv3

SMBv3 is a more complex protocol than a simple TCP file transfer, with additional overhead for headers, encryption, and other features like window scaling, pipelining, and multi-channel transfers. To adjust our calculations for SMBv3, we need to factor in this overhead. Let's break this down:

  1. Overhead for SMBv3

SMBv3 adds protocol headers to each packet. The typical overhead for SMBv3 can range from 60 to 100 bytes per packet depending on the specific implementation, encryption, and other factors. We’ll use a conservative estimate of 100 bytes of overhead per packet (this includes SMBv3, TCP, and IP headers).

Thus, the effective payload size of each packet will be reduced from the MTU size of 1470 bytes: Effective Payload=1470 bytes−100 bytes=1370 bytes Effective Payload=1470bytes−100bytes=1370bytes 2. Number of Packets Required (with Overhead)

Now we calculate how many packets are needed to transfer the 100 KB file using this reduced payload size: Number of Packets=File SizeEffective Payload=102,400 bytes1370 bytes/packet≈74.82 packets Number of Packets=Effective PayloadFile Size​=1370bytes/packet102,400bytes​≈74.82packets

Rounding up, we need 75 packets. 3. SMBv3 Overhead on Transfer Efficiency

SMBv3 also uses pipelining and window scaling to improve efficiency, reducing the need for an RTT for every packet. Let's assume a window size of 10 packets (which is typical for SMBv3 with good pipelining).

For every window, we send 10 packets and wait for 1 RTT for acknowledgment. 4. Total Number of RTTs

Now calculate the number of RTTs based on the window size: Total RTTs=Total PacketsWindow Size=7510=7.5 RTTs Total RTTs=Window SizeTotal Packets​=1075​=7.5RTTs

Since we can't have a fraction of an RTT, we round up to 8 RTTs. 5. Round-Trip Time (RTT)

We already know the RTT for this link is 40 ms or 0.04 seconds. 6. Total Transmission Time with SMBv3 Overhead

The total transmission time will be: Total Transmission Time=Total RTTs×RTT=8×0.04 seconds=0.32 seconds Total Transmission Time=Total RTTs×RTT=8×0.04seconds=0.32seconds 7. Additional Overhead and Processing Time

We’ll assume 10% additional time for other processing overheads in SMBv3 (e.g., encryption, decryption, and internal protocol processing).

Thus, the final expected time becomes: Final Expected Time=0.32 seconds×1.10=0.352 seconds Final Expected Time=0.32seconds×1.10=0.352seconds 8. Conclusion

The expected copy duration for a 100 KB file over a 1470-byte MTU, 20 ms delay link, using SMBv3 (with typical overhead) is approximately 0.352 seconds or 352 milliseconds.

This takes into account SMBv3 protocol overhead, pipelining, and processing time.

ETL as a service

Workflow as a service

Citrix Automated Reporting

Code to create automated citrix acceptance reports

import requests
import json
from requests.auth import HTTPBasicAuth

# Citrix VPX credentials and Nitro API endpoint
citrix_vpx_url = 'https://<citrix-vpx-ip>/nitro/v1/config/'
username = '<your-username>'
password = '<your-password>'
auth = HTTPBasicAuth(username, password)

# Disable SSL warnings for self-signed certificates
requests.packages.urllib3.disable_warnings()

# Function to get data from the Nitro API
def get_data_from_vpx(endpoint):
    url = citrix_vpx_url + endpoint
    response = requests.get(url, auth=auth, verify=False)
    if response.status_code == 200:
        return response.json()
    else:
        print(f"Failed to fetch {endpoint}: {response.status_code}")
        return None

# Fetch system-level details
def get_system_report():
    system_info = get_data_from_vpx('nsconfig')
    if system_info:
        ns_ip_address = system_info['nsconfig'][0].get('IPAddress', 'N/A')
        hostname = system_info['nsconfig'][0].get('hostname', 'N/A')
        ha_status = system_info['nsconfig'][0].get('hacurstatus', 'N/A')
        print(f"System IP Address: {ns_ip_address}")
        print(f"Hostname: {hostname}")
        print(f"HA Status: {ha_status}\n")

# Fetch and classify backend health status for virtual servers, also calculate health score
def get_load_balancer_report():
    lb_vservers = get_data_from_vpx('lbvserver')
    if lb_vservers:
        print(f"Load Balancing Virtual Servers:\n{'-'*30}")

        # Separate servers by health status
        perfect_servers = []
        degraded_servers = []
        down_servers = []
        
        # Counters for health score calculation
        total_servers = 0
        perfect_count = 0
        degraded_count = 0

        for vserver in lb_vservers['lbvserver']:
            name = vserver.get('name', 'N/A')
            ip = vserver.get('ipv46', 'N/A')
            state = vserver.get('curstate', 'N/A')
            total_servers += 1

            # Fetch backend service health status
            services_bound = get_data_from_vpx(f'lbvserver_service_binding/{name}')
            if services_bound:
                backend_health = {'perfect': True, 'degraded': False}
                for service in services_bound.get('lbvserver_service_binding', []):
                    service_name = service.get('servicename', 'N/A')
                    service_health = get_data_from_vpx(f'service/{service_name}')
                    if service_health:
                        health_status = service_health['service'][0].get('svrstate', 'N/A')
                        if health_status == 'DOWN':
                            backend_health['perfect'] = False
                            backend_health['degraded'] = True
                        elif health_status != 'UP':
                            backend_health['perfect'] = False

                if backend_health['perfect']:
                    perfect_servers.append(f"Name: {name} | IP: {ip} | State: {state}")
                    perfect_count += 1
                elif backend_health['degraded']:
                    degraded_servers.append(f"Name: {name} | IP: {ip} | State: {state}")
                    degraded_count += 1
                else:
                    down_servers.append(f"Name: {name} | IP: {ip} | State: {state}")
            else:
                down_servers.append(f"Name: {name} | IP: {ip} | State: {state}")

        # Print out categorized servers
        print("Perfect (Green) Servers:\n" + "-" * 30)
        for server in perfect_servers:
            print(server)
        print("\nDegraded (Yellow) Servers:\n" + "-" * 30)
        for server in degraded_servers:
            print(server)
        print("\nDown (Red) Servers:\n" + "-" * 30)
        for server in down_servers:
            print(server)
        print('\n')

        # Calculate overall health score
        total_degraded = degraded_count * 0.5
        total_health_contrib = perfect_count + total_degraded
        overall_health_score = (total_health_contrib / total_servers) * 100 if total_servers > 0 else 0

        print(f"Overall Health Score: {overall_health_score:.2f}%\n")

# Fetch SSL certificate details, including expiration dates
def get_ssl_report():
    ssl_certificates = get_data_from_vpx('sslcertkey')
    if ssl_certificates:
        print(f"SSL Certificates:\n{'-'*30}")
        for cert in ssl_certificates['sslcertkey']:
            cert_name = cert.get('certkey', 'N/A')
            cert_alias = cert.get('cert', 'N/A')
            exp_date = cert.get('expirymonitor', 'N/A')  # This field might vary depending on your Citrix version
            print(f"Cert Name: {cert_name} | Alias: {cert_alias} | Expiry Date: {exp_date}")
        print('\n')

# Fetch high availability details
def get_ha_report():
    ha_status = get_data_from_vpx('hanode')
    if ha_status:
        print(f"High Availability Details:\n{'-'*30}")
        for node in ha_status['hanode']:
            node_id = node.get('id', 'N/A')
            node_state = node.get('state', 'N/A')
            print(f"Node ID: {node_id} | State: {node_state}")
        print('\n')

# Main function to generate the report
def generate_report():
    print("Citrix VPX High-Level Report\n")
    get_system_report()
    get_load_balancer_report()
    get_ssl_report()
    get_ha_report()

if __name__ == '__main__':
    generate_report()

PBS Tests

Baseline

Delay

Data aşağıkdaki gibi

vm1 için

vm2 için

vm3 için

Hammerdb çalışırken

NE assesement

Layer Question Best Answer Explanation
1 - Physical What are the differences between single-mode and multi-mode fiber in terms of distance and bandwidth? Single-mode fiber (SMF) supports longer distances (up to 40 km) using a small core and a laser light source, while multi-mode fiber (MMF) is limited to shorter distances (typically up to 550m) using an LED light source. SMF is preferred for long-haul and metro networks, whereas MMF is used in short-range applications like data centers.
1 - Physical How does signal attenuation affect network performance, and how can you mitigate it in fiber and copper cabling? Signal attenuation weakens signals over distance. In copper, use repeaters or shorter cables. In fiber, use higher-quality optics and proper splicing. Attenuation leads to packet loss and slower speeds; fiber is less prone to interference than copper.
2 - Data Link What is the difference between VLANs and VXLANs, and how do they impact network segmentation? VLANs segment Layer 2 traffic within a local network, while VXLANs extend Layer 2 segments over Layer 3 using encapsulation (UDP port 4789). VXLANs allow scalability beyond traditional VLANs (4094 limit) and support multi-tenant cloud environments.
2 - Data Link Explain how Spanning Tree Protocol (STP) prevents network loops and describe one alternative protocol that can replace STP. STP detects and disables redundant links to prevent loops, using a root bridge election. Alternative: RSTP (Rapid STP) provides faster convergence, or TRILL/SPB eliminates the need for STP by using shortest-path forwarding. Without STP, broadcast storms can cripple networks. Modern alternatives improve convergence and efficiency.
3 - Network How does BGP determine the best route to a destination, and what factors can influence its decision? BGP selects the best path using attributes like AS-Path (shortest route wins), Local Preference (higher is better), MED (lower is preferred), and Next-Hop reachability. BGP is a path-vector protocol used for inter-domain routing and is critical for internet and ISP-level routing.
3 - Network Explain the difference between NAT, PAT, and how they impact IPv4 addressing in a corporate environment. NAT (Network Address Translation) maps private IPs to a public IP; PAT (Port Address Translation) allows multiple private IPs to share one public IP using port numbers. PAT is widely used in enterprises to conserve public IPs, enabling many devices to access the internet using a single IP.
4 - Transport Compare and contrast TCP and UDP. In what scenarios would you choose one over the other? TCP is reliable (error checking, retransmissions) and used for HTTP, SSH, etc. UDP is faster but unreliable, used for VoIP, DNS, and video streaming. TCP is preferred for critical applications, while UDP is used when low latency is required.
4 - Transport Explain the significance of the TCP three-way handshake and how it impacts security vulnerabilities such as SYN floods. The handshake (SYN → SYN-ACK → ACK) establishes a connection. SYN flood attacks exploit this by sending repeated SYNs without completing the handshake. Firewalls and SYN cookies can help mitigate SYN flood attacks by preventing resource exhaustion.
5 - Session How does SSL/TLS establish a secure session between a client and a server? It uses asymmetric encryption for key exchange (handshake) and symmetric encryption for data transfer (AES, ChaCha20). TLS ensures confidentiality, integrity, and authentication using certificates (X.509).
5 - Session What is the role of session persistence (sticky sessions) in load balancing, and how does it affect performance? It ensures requests from the same client go to the same backend server, improving user experience. While helpful for stateful applications, it can cause uneven load distribution if not managed properly.
6 - Presentation How do encryption and compression impact network performance at the presentation layer? Encryption secures data but increases CPU usage. Compression reduces data size, improving transmission speeds. Balancing security (encryption) and efficiency (compression) is key in optimizing network performance.
6 - Presentation Explain the difference between Base64 encoding and actual encryption. In what scenarios would you use each? Base64 encodes data for safe transport but is reversible. Encryption (AES, RSA, etc.) secures data by making it unreadable without a key. Use Base64 for data encoding (e.g., email attachments). Use encryption for security (e.g., passwords, confidential data).
7 - Application What are the differences between HTTP/1.1, HTTP/2, and HTTP/3 in terms of performance and security? HTTP/1.1 uses sequential requests; HTTP/2 introduces multiplexing and header compression; HTTP/3 uses QUIC for lower latency and better security. HTTP/3 improves web performance by reducing round-trip delays and mitigating TCP head-of-line blocking.
7 - Application How does DNS caching work, and what are the security risks associated with DNS cache poisoning? DNS caching stores query results locally to speed up lookups. Cache poisoning injects malicious IP mappings into caches, redirecting users to fraudulent sites. Mitigations include DNSSEC (signing records) and validating responses to prevent tampering.
7 - Application What is the role of an API gateway in modern cloud networking, and how does it impact microservices architectures? It manages API traffic, enforcing authentication, rate limiting, and load balancing for microservices. API gateways help decouple services, improve security, and enable scalable architectures.

System Requirements Specification (SRS)

Project: Layer 2 Data Center Extender (L2 DCE)

Version: 1.0
Date: 29/04/2025
Author: Mesut Bayrak

1. Introduction

1.1 Purpose

Describe the purpose of the L2 DCE system (e.g., extend VLANs across data centers with low latency).

1.2 Scope

Define what the L2 DCE will and will not do.

1.3 Definitions, Acronyms, and Abbreviations

1.4 References

List applicable RFCs, whitepapers, standards.

1.5 Overview

Brief description of document structure.

2. Overall Description

Main purpose of the device is to provide a transparent Layer-2 connectivity between two separate sites.

2.1 Product Perspective

Device can have two distinct operational units to be located in two sites. One central unit (an aggragator), and one (or more) remote site unit(s).

Central unit is intended to be located in a data-center facility with proper power and cooling.

Remote side units are intended to be located inside out-of-reach facilities, via some rental or sale agreement.

Describe how L2 DCE fits into the current system (optional: diagram).

2.2 Product Functions

Function Definition:

Operational Definition:

2.3 User Characteristics

2.4 Constraints

2.5 Assumptions and Dependencies

3. Specific Requirements

3.1 Functional Requirements

ID Requirement Description
FR-1 VLAN Extension Extend VLANs 1–4094.
FR-3 Control Plane Use BGP EVPN or similar.
FR-4 Failover To be defined in future
FR-5 MTU Handling Minimum MTU 9100 bytes.

3.2 Non-Functional Requirements

ID Requirement Description
NFR-1 Availability 99.99% uptime.
NFR-3 Performance Latency increase < 2ms one-way.

4. Interface Requirements

4.1 Hardware Interfaces

4.2 Software Interfaces

4.3 Communication Interfaces

5. Performance Requirements

6. Security Requirements

7. Other Requirements

8. Appendix

extender Daemon